From kmgraph
Use when a bug, failing test, blocked plan task, integration failure, or reproducibility issue has resisted 3+ distinct fix attempts or 30+ minutes of effort. Activates powerful-tier diagnosis, enforces hypothesis logging, and drives to a structured exit-path decision.
npx claudepluginhub technomensch/knowledge-graph --plugin kmgraphThis skill uses the workspace's default tool permissions.
Before starting, read `~/.kmgraph/rules.md` — Plan Protocol § Stuck-Work Escalation for thresholds, scope rules, and exit paths.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Retrieves current documentation, API references, and code examples for libraries, frameworks, SDKs, CLIs, and services via Context7 CLI. Ideal for API syntax, configs, migrations, and setup queries.
Uses ctx7 CLI to fetch current library docs, manage AI coding skills (install/search/generate), and configure Context7 MCP for AI editors.
Before starting, read ~/.kmgraph/rules.md — Plan Protocol § Stuck-Work Escalation for thresholds, scope rules, and exit paths.
Trigger when work has a definable success criterion and has resisted resolution. Applies to:
Does not apply to exploratory or iterative work (prompt tuning, CSS iteration, prototype spiking).
Tier note: This skill declares
required_tier: powerful-tier. It does not collapse to a lower tier — if powerful-tier is unavailable, the skill halts with an explicit error rather than running on a lower-capability model.
Create meta-issue (if not already open):
/kmgraph:meta-issue "[Problem Title]"
Compile attempt log — read all prior attempts from meta-issue or conversation. For each attempt, confirm a distinct hypothesis was tested (not a retry of the same fix).
powerful-tier diagnosis — provide powerful-tier with:
root-cause-evolution.md if meta-issue exists
powerful-tier reviews and provides fresh diagnosis and next hypothesis.Log next attempt with powerful-tier-recommended hypothesis:
/kmgraph:meta-issue --log-attempt NNN "[hypothesis]"
powerful-tier involvement continues — powerful-tier reviews each subsequent attempt result before the next attempt begins. Maximum 3 powerful-tier rounds before forcing exit-path analysis (step below).
Present the following to the user and require a decision:
| Question | Answer |
|---|---|
| Would a brainstorming session reframe this? | Yes / No |
| Is there a known workaround? | [describe or No] |
| Is this required functionality or nice-to-have? | Required / Nice-to-have |
| Will the system break without it? | Yes / No / Partially |
| Will users be impacted? | Yes / No / [describe] |
| Can this be deferred to a future release? | Yes / No |
Then select a mandatory exit path:
Present the exit-path analysis and selected path following plan approval rules. Wait for explicit approval before proceeding.
If root cause genuinely shifts (new diagnosis invalidates prior attempts), reset the attempt counter. Log the reset in analysis/root-cause-evolution.md with reasoning.