diff --git a/plugins/aidd-context/skills/00-onboard/SKILL.md b/plugins/aidd-context/skills/00-onboard/SKILL.md
index 443914f1c..3dcbd80d1 100644
--- a/plugins/aidd-context/skills/00-onboard/SKILL.md
+++ b/plugins/aidd-context/skills/00-onboard/SKILL.md
@@ -30,3 +30,17 @@ Run the actions in that order, looping. Read an action's file in `actions/` befo
 - Never test a plugin version against a registry.
 - Never trust a stale status.
 - Wait for an explicit reply before running anything.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:00-onboard"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/01-bootstrap/SKILL.md b/plugins/aidd-context/skills/01-bootstrap/SKILL.md
index e4db4a84a..7f462b7a2 100644
--- a/plugins/aidd-context/skills/01-bootstrap/SKILL.md
+++ b/plugins/aidd-context/skills/01-bootstrap/SKILL.md
@@ -37,3 +37,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 
 - `assets/checklist.md` - the 24-item checklist (4 blocks)
 - `assets/install-template.md` - the `INSTALL.md` skeleton
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:01-bootstrap"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/02-project-memory/SKILL.md b/plugins/aidd-context/skills/02-project-memory/SKILL.md
index e854cb94b..bf54598c1 100644
--- a/plugins/aidd-context/skills/02-project-memory/SKILL.md
+++ b/plugins/aidd-context/skills/02-project-memory/SKILL.md
@@ -31,3 +31,17 @@ Run the flow above, reading only the next action file.
 - Ask before anything ambiguous. Never default silently.
 - A bank that already exists changes only through what the user approved, file by file and line by line.
 - End with a short report of what changed.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:02-project-memory"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/03-context-generate/SKILL.md b/plugins/aidd-context/skills/03-context-generate/SKILL.md
index 6040a4dfa..0fdfb4586 100644
--- a/plugins/aidd-context/skills/03-context-generate/SKILL.md
+++ b/plugins/aidd-context/skills/03-context-generate/SKILL.md
@@ -19,3 +19,17 @@ Routes a generation request to the dedicated generator for the artifact kind. Ho
 | hook     | `aidd-context:08-hook-generate`  |
 
 Identify the artifact kind from the request, then hand off to the matching generator. If the kind is unclear, ask which one. To survey or list existing artifacts, use the explore skill instead.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:03-context-generate"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/04-skill-generate/SKILL.md b/plugins/aidd-context/skills/04-skill-generate/SKILL.md
index 6e546fb72..fc0d3fede 100644
--- a/plugins/aidd-context/skills/04-skill-generate/SKILL.md
+++ b/plugins/aidd-context/skills/04-skill-generate/SKILL.md
@@ -29,3 +29,17 @@ Run the flow above. Read only the next action file.
 - If a cited reference cannot be read, stop and report the missing file.
 - Confirm every target and name with the user.
 - Never write silently.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:04-skill-generate"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/05-rule-generate/SKILL.md b/plugins/aidd-context/skills/05-rule-generate/SKILL.md
index f23e0a26b..66e698d93 100644
--- a/plugins/aidd-context/skills/05-rule-generate/SKILL.md
+++ b/plugins/aidd-context/skills/05-rule-generate/SKILL.md
@@ -27,3 +27,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/rule-template.md`: rule file scaffold.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:05-rule-generate"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/06-agent-generate/SKILL.md b/plugins/aidd-context/skills/06-agent-generate/SKILL.md
index aac9ebe55..062973e12 100644
--- a/plugins/aidd-context/skills/06-agent-generate/SKILL.md
+++ b/plugins/aidd-context/skills/06-agent-generate/SKILL.md
@@ -27,3 +27,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/agent-template.md`: agent file scaffold.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:06-agent-generate"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/07-command-generate/SKILL.md b/plugins/aidd-context/skills/07-command-generate/SKILL.md
index b986cf964..1ebcbe86f 100644
--- a/plugins/aidd-context/skills/07-command-generate/SKILL.md
+++ b/plugins/aidd-context/skills/07-command-generate/SKILL.md
@@ -27,3 +27,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/command-template.md`: command file scaffold.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:07-command-generate"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/08-hook-generate/SKILL.md b/plugins/aidd-context/skills/08-hook-generate/SKILL.md
index 5cd7a8838..acd439ee8 100644
--- a/plugins/aidd-context/skills/08-hook-generate/SKILL.md
+++ b/plugins/aidd-context/skills/08-hook-generate/SKILL.md
@@ -28,3 +28,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 
 - `assets/hook-template.json`: the entry scaffold.
 - `assets/hook-script-template.sh`: the backing-script scaffold.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:08-hook-generate"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/09-mermaid/SKILL.md b/plugins/aidd-context/skills/09-mermaid/SKILL.md
index 34f74dad8..bd99cc66a 100644
--- a/plugins/aidd-context/skills/09-mermaid/SKILL.md
+++ b/plugins/aidd-context/skills/09-mermaid/SKILL.md
@@ -27,3 +27,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## References
 
 - `references/mermaid-conventions.md`: the conventions and defaults every generated diagram follows.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:09-mermaid"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/10-learn/SKILL.md b/plugins/aidd-context/skills/10-learn/SKILL.md
index 8e1c5e63c..a7e5e3344 100644
--- a/plugins/aidd-context/skills/10-learn/SKILL.md
+++ b/plugins/aidd-context/skills/10-learn/SKILL.md
@@ -33,3 +33,17 @@ Run the flow above. Read only the next action file.
 - Write only the user-approved plan.
 - Preserve user edits and touch affected files only.
 - Write project files only, never personal or global memory.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:10-learn"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/11-explore/SKILL.md b/plugins/aidd-context/skills/11-explore/SKILL.md
index 48fbae879..2cc27ecea 100644
--- a/plugins/aidd-context/skills/11-explore/SKILL.md
+++ b/plugins/aidd-context/skills/11-explore/SKILL.md
@@ -37,3 +37,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## References
 
 - `references/ai-mapping.md`: the per-tool signals, scan paths, and formats for the Tooling and Context axes.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:11-explore"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-context/skills/12-cook/SKILL.md b/plugins/aidd-context/skills/12-cook/SKILL.md
index e9f92d63e..b0bd1e82f 100644
--- a/plugins/aidd-context/skills/12-cook/SKILL.md
+++ b/plugins/aidd-context/skills/12-cook/SKILL.md
@@ -29,3 +29,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 
 - `assets/recipe-template.md`: the canonical recipe scaffold `upsert` renders from, and the shape `list` parses.
 - `assets/recipes/`: bundled recipes shipped with this skill.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-context:12-cook"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/02-implement/SKILL.md b/plugins/aidd-dev/skills/02-implement/SKILL.md
index dd77f0fd3..42174a99b 100644
--- a/plugins/aidd-dev/skills/02-implement/SKILL.md
+++ b/plugins/aidd-dev/skills/02-implement/SKILL.md
@@ -27,3 +27,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## References
 
 - `references/blocked.md`: the conditions that make a plan `blocked` and need a human.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:02-implement"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/03-assert/SKILL.md b/plugins/aidd-dev/skills/03-assert/SKILL.md
index 2d94635cf..d57588886 100644
--- a/plugins/aidd-dev/skills/03-assert/SKILL.md
+++ b/plugins/aidd-dev/skills/03-assert/SKILL.md
@@ -29,3 +29,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/task-template.md`: the tracking file the frontend facet fills across its fix loop.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:03-assert"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/04-audit/SKILL.md b/plugins/aidd-dev/skills/04-audit/SKILL.md
index 768ba5ec8..fed52e601 100644
--- a/plugins/aidd-dev/skills/04-audit/SKILL.md
+++ b/plugins/aidd-dev/skills/04-audit/SKILL.md
@@ -35,3 +35,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/audit-template.md`: the report structure both run modes fill.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:04-audit"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/05-review/SKILL.md b/plugins/aidd-dev/skills/05-review/SKILL.md
index 4da8ad889..20f800896 100644
--- a/plugins/aidd-dev/skills/05-review/SKILL.md
+++ b/plugins/aidd-dev/skills/05-review/SKILL.md
@@ -38,3 +38,17 @@ Run all three by default, composing one report. Run a single axis only when the
 
 - `assets/review-template.md`: the single report the three axes fill.
 - `assets/review-validator.yml`: the closed set of report sections.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:05-review"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/06-test/SKILL.md b/plugins/aidd-dev/skills/06-test/SKILL.md
index 8500963e4..9687fdee2 100644
--- a/plugins/aidd-dev/skills/06-test/SKILL.md
+++ b/plugins/aidd-dev/skills/06-test/SKILL.md
@@ -24,3 +24,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 - One action per run: follow only the matching action file.
 - Functional behavior only: never couple a test to implementation details.
 - Never compromise quality for speed.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:06-test"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/07-refactor/SKILL.md b/plugins/aidd-dev/skills/07-refactor/SKILL.md
index bcfabc6fd..a5c48b10c 100644
--- a/plugins/aidd-dev/skills/07-refactor/SKILL.md
+++ b/plugins/aidd-dev/skills/07-refactor/SKILL.md
@@ -27,3 +27,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 - Audit-fed, optional: when the caller pushes an audit report (a path under `aidd_docs/tasks/audits/` or pasted findings), take its findings for this axis as the fix list and skip the scan. The bridge is the report artifact; this skill never loads or calls another skill. The audit `code-quality` pillar feeds the `cleanup` axis; the other axes map by name.
 - Severity uses the shared 3-level scale: 🔴 critical, 🟡 warning, 🟢 minor.
 - Stay inside the axis: dependency upgrades and UI redesign are out of scope. Add tests only as a regression for a security fix, never otherwise.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:07-refactor"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/08-debug/SKILL.md b/plugins/aidd-dev/skills/08-debug/SKILL.md
index 795597e68..44c724968 100644
--- a/plugins/aidd-dev/skills/08-debug/SKILL.md
+++ b/plugins/aidd-dev/skills/08-debug/SKILL.md
@@ -32,3 +32,17 @@ Pick the one action matching the intent; never default to `01`. Triggers like "r
 ## References
 
 - `references/mermaid-conventions.md`: the project's Mermaid conventions for the action-path flowchart.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:08-debug"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/09-for-sure/SKILL.md b/plugins/aidd-dev/skills/09-for-sure/SKILL.md
index 31d5be3e9..80e39de48 100644
--- a/plugins/aidd-dev/skills/09-for-sure/SKILL.md
+++ b/plugins/aidd-dev/skills/09-for-sure/SKILL.md
@@ -35,3 +35,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## References
 
 - `references/autonomous-loop-log-format.md`: the Log entry format the loop appends per attempt.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:09-for-sure"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/10-todo/SKILL.md b/plugins/aidd-dev/skills/10-todo/SKILL.md
index a2083c643..63d5f603d 100644
--- a/plugins/aidd-dev/skills/10-todo/SKILL.md
+++ b/plugins/aidd-dev/skills/10-todo/SKILL.md
@@ -15,3 +15,17 @@ actions/01-todo.md
 ```
 
 Before running an action, read its file in `actions/`, not only the table or assets.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:10-todo"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-dev/skills/11-browser-qa/SKILL.md b/plugins/aidd-dev/skills/11-browser-qa/SKILL.md
index d886556e5..38f51e91f 100644
--- a/plugins/aidd-dev/skills/11-browser-qa/SKILL.md
+++ b/plugins/aidd-dev/skills/11-browser-qa/SKILL.md
@@ -27,3 +27,17 @@ Read only the next action's file before running it.
 - Run against a reviewed change and never patch the application.
 - Never spawn agents. Batch independent reads and tool checks, but keep state-changing browser work sequential.
 - Do not narrate action transitions, searches, fixtures, selectors, or successful checks. Report only a blocker, a required decision, or the final verdict and paths.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-dev:11-browser-qa"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/01-ticket-info/SKILL.md b/plugins/aidd-pm/skills/01-ticket-info/SKILL.md
index 1e90b79b1..b186eb641 100644
--- a/plugins/aidd-pm/skills/01-ticket-info/SKILL.md
+++ b/plugins/aidd-pm/skills/01-ticket-info/SKILL.md
@@ -18,3 +18,17 @@ Run the flow above. Read only the next action file.
 | Action        | Does                                                     |
 | ------------- | --------------------------------------------------------- |
 | ticket-info   | resolve ticket id, query configured tool, display fields |
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:01-ticket-info"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/02-user-stories/SKILL.md b/plugins/aidd-pm/skills/02-user-stories/SKILL.md
index 007811660..c5d779960 100644
--- a/plugins/aidd-pm/skills/02-user-stories/SKILL.md
+++ b/plugins/aidd-pm/skills/02-user-stories/SKILL.md
@@ -39,3 +39,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - Draft only after actor, need, and outcome are explicit in the source or confirmed.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:02-user-stories"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/03-prd/SKILL.md b/plugins/aidd-pm/skills/03-prd/SKILL.md
index 683b1ba32..3a4bbca6d 100644
--- a/plugins/aidd-pm/skills/03-prd/SKILL.md
+++ b/plugins/aidd-pm/skills/03-prd/SKILL.md
@@ -28,3 +28,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - State what and why; never a technical plan or user stories.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:03-prd"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/04-spec/SKILL.md b/plugins/aidd-pm/skills/04-spec/SKILL.md
index 10afd34c0..b63ffe5d6 100644
--- a/plugins/aidd-pm/skills/04-spec/SKILL.md
+++ b/plugins/aidd-pm/skills/04-spec/SKILL.md
@@ -28,3 +28,17 @@ Run the flow above. Read only the next action file.
 - Hold intent, never implementation: solution-agnostic, no how, few acceptance criteria.
 - Keep it readable: clear headers, bulleted criteria, explicit non-goals.
 - Immutable once validated: never rewrite a locked spec.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:04-spec"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/05-spike/SKILL.md b/plugins/aidd-pm/skills/05-spike/SKILL.md
index b64bf12d5..e43050a58 100644
--- a/plugins/aidd-pm/skills/05-spike/SKILL.md
+++ b/plugins/aidd-pm/skills/05-spike/SKILL.md
@@ -34,3 +34,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - Bound the question; never answer beyond its stop condition.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:05-spike"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/06-product-brief/SKILL.md b/plugins/aidd-pm/skills/06-product-brief/SKILL.md
index 0decc9c02..1970e0855 100644
--- a/plugins/aidd-pm/skills/06-product-brief/SKILL.md
+++ b/plugins/aidd-pm/skills/06-product-brief/SKILL.md
@@ -41,3 +41,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - Frame the opportunity; never write requirements.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:06-product-brief"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/07-epic/SKILL.md b/plugins/aidd-pm/skills/07-epic/SKILL.md
index 299be569e..c8dd0aec7 100644
--- a/plugins/aidd-pm/skills/07-epic/SKILL.md
+++ b/plugins/aidd-pm/skills/07-epic/SKILL.md
@@ -33,3 +33,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - Stay at outcome level; do not propose child slices or implementation.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:07-epic"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/08-three-amigos/SKILL.md b/plugins/aidd-pm/skills/08-three-amigos/SKILL.md
index 077bf3338..23faad118 100644
--- a/plugins/aidd-pm/skills/08-three-amigos/SKILL.md
+++ b/plugins/aidd-pm/skills/08-three-amigos/SKILL.md
@@ -28,3 +28,17 @@ Run the flow above. Read only the next action file.
 - Treat roles as analytical lenses, not human authorities.
 - Ground every finding and question in cited evidence; never invent a decision.
 - Return proposals only; the caller decides what to do with them.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:08-three-amigos"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/09-defect/SKILL.md b/plugins/aidd-pm/skills/09-defect/SKILL.md
index 23300d1a3..e215bc564 100644
--- a/plugins/aidd-pm/skills/09-defect/SKILL.md
+++ b/plugins/aidd-pm/skills/09-defect/SKILL.md
@@ -33,3 +33,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - Record the mismatch; never diagnose or implement its fix.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:09-defect"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-pm/skills/10-task/SKILL.md b/plugins/aidd-pm/skills/10-task/SKILL.md
index fccffb895..d69a113f4 100644
--- a/plugins/aidd-pm/skills/10-task/SKILL.md
+++ b/plugins/aidd-pm/skills/10-task/SKILL.md
@@ -33,3 +33,17 @@ Run the flow above. Read only the next action file.
 - Ask natural questions; never expose actions, references, or unchanged state.
 - Require explicit approval or caller-provided bounded authority before any write.
 - Record delivery work; never plan or implement it.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-pm:10-task"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-refine/skills/01-brainstorm/SKILL.md b/plugins/aidd-refine/skills/01-brainstorm/SKILL.md
index e0eb39c62..975321339 100644
--- a/plugins/aidd-refine/skills/01-brainstorm/SKILL.md
+++ b/plugins/aidd-refine/skills/01-brainstorm/SKILL.md
@@ -32,3 +32,17 @@ Run the flow above. Read only the next action's file before running it.
 - State a leaning and its tradeoff when facts already point one way.
 - Hide process words: no density, coverage, nodes, completeness, matrix, or frame unless the user asks for an audit.
 - Wait after questions, approval, and persistence choices.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-refine:01-brainstorm"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-refine/skills/02-challenge/SKILL.md b/plugins/aidd-refine/skills/02-challenge/SKILL.md
index 4e1a03c58..4ed9ad5ff 100644
--- a/plugins/aidd-refine/skills/02-challenge/SKILL.md
+++ b/plugins/aidd-refine/skills/02-challenge/SKILL.md
@@ -29,3 +29,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/report-template.md`: findings report skeleton, filled per run.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-refine:02-challenge"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-refine/skills/03-shadow-areas/SKILL.md b/plugins/aidd-refine/skills/03-shadow-areas/SKILL.md
index a09d1fd55..88a1dd33b 100644
--- a/plugins/aidd-refine/skills/03-shadow-areas/SKILL.md
+++ b/plugins/aidd-refine/skills/03-shadow-areas/SKILL.md
@@ -38,3 +38,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/report-template.md`: report skeleton with header, per-category sections, and `status: clean` block.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-refine:03-shadow-areas"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-refine/skills/04-fact-check/SKILL.md b/plugins/aidd-refine/skills/04-fact-check/SKILL.md
index debeeb30d..786d81cdf 100644
--- a/plugins/aidd-refine/skills/04-fact-check/SKILL.md
+++ b/plugins/aidd-refine/skills/04-fact-check/SKILL.md
@@ -39,3 +39,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/report-template.md`: rewritten-answer skeleton with the `## Sources` footnote block.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-refine:04-fact-check"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-telemetry/skills/00-init/SKILL.md b/plugins/aidd-telemetry/skills/00-init/SKILL.md
index 023d8d121..17e41c4f3 100644
--- a/plugins/aidd-telemetry/skills/00-init/SKILL.md
+++ b/plugins/aidd-telemetry/skills/00-init/SKILL.md
@@ -55,3 +55,17 @@ Run the flow above. Read only the next action file.
 - Removing what was measured is irreversible, and reaches further than one project: this machine's stored records span every project ever measured on it, not only the one being discussed. Never presented as reversible, and never presented as scoped to one project when it is not.
 - What history keeps is relayed exactly as `aidd telemetry forget` states it — a journal already committed as certainly held, one merely staged but never committed as not yet held, one not tracked at all as possibly held — and is never presented as removable. No command in this skill rewrites git history.
 - The `aidd` command cannot be found: say so and change nothing.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-telemetry:00-init"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-telemetry/skills/01-cost/SKILL.md b/plugins/aidd-telemetry/skills/01-cost/SKILL.md
index 4892ae2b2..6e9302210 100644
--- a/plugins/aidd-telemetry/skills/01-cost/SKILL.md
+++ b/plugins/aidd-telemetry/skills/01-cost/SKILL.md
@@ -50,3 +50,17 @@ menu and ask them to pick.
 - An absent number is not a zero. Say the figure is unknown and give what is known instead.
 - Turning measurement on belongs elsewhere. Stop and say so rather than doing it here.
 - The script cannot be found or fails: say so and show no figure.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-telemetry:01-cost"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-telemetry/skills/02-check/SKILL.md b/plugins/aidd-telemetry/skills/02-check/SKILL.md
index 514839d55..52b3ef0ec 100644
--- a/plugins/aidd-telemetry/skills/02-check/SKILL.md
+++ b/plugins/aidd-telemetry/skills/02-check/SKILL.md
@@ -33,3 +33,17 @@ Run the flow above. Read only the next action file.
 - `ok`, `FAIL` and `--` are three different answers. `--` means there was nothing to evaluate, not that the chain is healthy.
 - A declaration that the recorder is set up is not proof it fired. Relay it as what it is — a fact about configuration, never a promise about behaviour.
 - The `aidd` command cannot be found: say so and check nothing.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-telemetry:02-check"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-ui/skills/01-hello/SKILL.md b/plugins/aidd-ui/skills/01-hello/SKILL.md
index e3406a95d..444150bda 100644
--- a/plugins/aidd-ui/skills/01-hello/SKILL.md
+++ b/plugins/aidd-ui/skills/01-hello/SKILL.md
@@ -19,3 +19,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Prerequisites
 
 - The plugin loaded locally (`claude --plugin-dir plugins/aidd-ui`, or installed from the marketplace).
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-ui:01-hello"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-vcs/skills/00-repo-init/SKILL.md b/plugins/aidd-vcs/skills/00-repo-init/SKILL.md
index 8c15d4119..90244939f 100644
--- a/plugins/aidd-vcs/skills/00-repo-init/SKILL.md
+++ b/plugins/aidd-vcs/skills/00-repo-init/SKILL.md
@@ -34,3 +34,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/CONTRIBUTING.md`: the project-root `CONTRIBUTING.md` template.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-vcs:00-repo-init"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-vcs/skills/01-commit/SKILL.md b/plugins/aidd-vcs/skills/01-commit/SKILL.md
index ea1286d44..8d722c9de 100644
--- a/plugins/aidd-vcs/skills/01-commit/SKILL.md
+++ b/plugins/aidd-vcs/skills/01-commit/SKILL.md
@@ -32,3 +32,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/commit-template.md`: Conventional commit format reference.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-vcs:01-commit"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-vcs/skills/02-pull-request/SKILL.md b/plugins/aidd-vcs/skills/02-pull-request/SKILL.md
index 82c212715..962a5dd6f 100644
--- a/plugins/aidd-vcs/skills/02-pull-request/SKILL.md
+++ b/plugins/aidd-vcs/skills/02-pull-request/SKILL.md
@@ -31,3 +31,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 
 - `assets/pull_request.md`: Request body template.
 - `assets/branch.md`: Branch-naming convention, the fallback when project memory sets none.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-vcs:02-pull-request"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-vcs/skills/03-release-tag/SKILL.md b/plugins/aidd-vcs/skills/03-release-tag/SKILL.md
index 9f7426a05..107e6628e 100644
--- a/plugins/aidd-vcs/skills/03-release-tag/SKILL.md
+++ b/plugins/aidd-vcs/skills/03-release-tag/SKILL.md
@@ -28,3 +28,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 ## Assets
 
 - `assets/release-template.md`: Release notes template.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-vcs:03-release-tag"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/plugins/aidd-vcs/skills/04-issue-create/SKILL.md b/plugins/aidd-vcs/skills/04-issue-create/SKILL.md
index 30c13a072..2e8b98abf 100644
--- a/plugins/aidd-vcs/skills/04-issue-create/SKILL.md
+++ b/plugins/aidd-vcs/skills/04-issue-create/SKILL.md
@@ -29,3 +29,17 @@ Before running an action, read its file in `actions/`, not only the table or ass
 
 - `assets/issue-template.md`: Issue / ticket body template.
 - `assets/CONTRIBUTING.md`: Contribution guidelines, including issue process.
+
+## Say when this skill's work is done
+
+Once this skill has produced what it was called for, and only then, run:
+
+```shell
+echo "aidd:step-end aidd-vcs:04-issue-create"
+```
+
+No host reports when a skill's work finished. A skill call's own result comes back in a
+tenth of a second, which is the dispatch and not the completion, so a measurement that
+never hears this ends the step where the next one begins — or, where none follows, at the
+journal's own last witnessed moment, which credits this skill with everything the session
+did afterwards.
diff --git a/scripts/__tests__/aidd-telemetry-step-end.test.js b/scripts/__tests__/aidd-telemetry-step-end.test.js
index e489d2a75..d41f8786b 100644
--- a/scripts/__tests__/aidd-telemetry-step-end.test.js
+++ b/scripts/__tests__/aidd-telemetry-step-end.test.js
@@ -128,3 +128,44 @@ test("every skill that opens a flow also says when that flow is over", () => {
     assert.equal(read, skill, `${skill} declares no end the hook reads as its own`);
   }
 });
+
+// Every skill, not only the ones that opted in. A skill that never says it is done leaves an
+// interval nothing closes: it runs to the next `step_start` or, failing that, to the journal's
+// own last witnessed moment, and everything the session did afterwards reads as that skill's
+// work. Measured on the one orchestrated session captured, 2026-09-04, before the three
+// orchestrators declared theirs: `aidd-dev:01-plan` opened at 06:00:50, closed nothing, and
+// took every one of the 972 records written over the three and a half hours that followed.
+//
+// A sweep and not a list, for the reason the sweep above already gives: a skill added to this
+// tree is covered without this file being told.
+function everySkillFile() {
+  const found = [];
+  const walk = (dir) => {
+    for (const entry of fs.readdirSync(dir, { withFileTypes: true })) {
+      const full = path.join(dir, entry.name);
+      if (entry.isDirectory()) walk(full);
+      else if (entry.name === "SKILL.md") found.push(full);
+    }
+  };
+  walk(pluginsDir);
+  return found;
+}
+
+test("every skill says when its own work is done", () => {
+  const silent = [];
+
+  for (const full of everySkillFile()) {
+    const relative = path.relative(pluginsDir, full);
+    const [plugin, , skillDir] = relative.split(path.sep);
+    const read = declaredStepEnd({
+      tool_input: { command: fs.readFileSync(full, "utf8") },
+    });
+    if (read !== `${plugin}:${skillDir}`) silent.push(`${relative} declares ${read}`);
+  }
+
+  assert.deepEqual(
+    silent,
+    [],
+    `A skill that never declares its end leaves an interval nothing closes.\n${silent.join("\n")}`
+  );
+});