Install
$ agentstack add skill-wfukatsu-nexus-architect-adapt-change ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Adapt to Change (Re-propagation Engine)
Desired Outcome
Apply a change to an existing product design with minimal, reversible re-runs:
- Change log —
reports/05_adaptation/change-log.md: the change (description,--type,
timestamp) and a before/after diff summary for every re-run artifact (for reversibility).
- Impact analysis —
reports/05_adaptation/impact-analysis.md: "change → impacted ID →
re-evaluate? + reason", i.e. the candidate blast radius after the judgment pass.
- Updated artifacts — the affected skills re-run with existing artifacts as input, and the
corresponding edges in work/traceability.json updated.
Invocation
/product:adapt-change --change="" [--type=constraint|market|competitor|tech|regulation] [--auto] [--lang=ja|en]
| Argument/Flag | Required | Description | |---------------|----------|-------------| | --change="" | Required | What changed (free text) | | --type=... | Recommended | Where the change enters the graph: constraint / market / competitor / tech / regulation | | --auto | Optional | Skip the human confirmation step (apply the judged impact set directly) | | --lang | Optional | Override output language |
Decision Criteria
- Minimal re-run. Never touch a skill the change does not reach. The graph proposes the
candidate set; the judgment pass decides whether each upstream reference still holds.
- Deterministic candidates, judged scope. Step 2 (transitive closure) is pure graph work;
step 3 (opus) expands/shrinks it with recorded reasons.
- Reversibility. Record a before/after diff summary for every rewritten artifact.
- Human checkpoint. Confirm the impact set before rewriting anything (unless
--auto). - Stop condition: change logged, impact set judged and confirmed, affected skills re-run,
traceability.json updated, and review run for coherence.
Prerequisites
| Input | Required/Recommended | Source | If missing/empty | |-------|---------------------|--------|------------------| | --change | Required | User | block with a message — nothing to propagate | | work/traceability.json | Required | all prior skills | block with a message — the engine reads only this | | existing artifacts | Required | prior skills | block — there must be a design to adapt |
Process
- Intake — record the change in
change-log.md(description,--type, timestamp passed in).
Apply @rules/product/adaptation-engine.md.
- Candidate blast radius (deterministic) — seed from the
--typehint, then walk
traceability.json upstream edges in reverse to get the downstream transitive closure.
- Judgment pass (opus) — for each candidate decide if its upstream reference still holds;
expand/shrink the set; write "change → impacted ID → re-evaluate? + reason" to impact-analysis.md.
- Confirm — present the impact set via
AskUserQuestion(skip under--auto). - Minimal re-run — re-run only the confirmed affected skills with existing artifacts as input;
record before/after diffs; update the affected edges in traceability.json.
- Coherence check — invoke
/product:review(consistency + traceability lenses) to catch
contradictions introduced by the re-propagation.
- Record — finalize both files; append the change summary to
work/context.md.
Output
reports/05_adaptation/change-log.md, reports/05_adaptation/impact-analysis.md, the re-run artifacts, and an updated work/traceability.json.
Reference Materials
| Resource | Purpose | |----------|---------| | @rules/product/adaptation-engine.md | Edge store, transitive-closure + judgment algorithm, principles |
Related Skills
| Skill | Relationship | |-------|-------------| | /product:review | Downstream — runs the coherence check after re-propagation | | /product:validate-assumptions | Related — a change may re-open the gate | | any pipeline skill | Re-invoked selectively as the confirmed impact set requires |
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: wfukatsu
- Source: wfukatsu/nexus-architect
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.