agentleFS
lenar-amirovlenar-amirov/product-pipeline-public/.claude/skills/next-advisor/SKILL.md

next-advisor

Skill13 starsChanged 5 months agoMITView on GitHub

Diagnoses the initiative state and recommends the 1-3 most valuable next actions. Use when the PM says "what's next", "что дальше", "continue", "где мы", "продолжаем", "what should I do", or seems unsure where to go — instead of mechanically suggesting the next pipeline step number.

The file

Files
SKILL.md

Next Advisor — /next

Recommend the most valuable next action from state, not from step numbers. "You are on step 6, next is step 7" is exactly what this skill replaces.

Inputs (read all that exist)

  1. output/hypotheses.json — the registry (or hypotheses.py show <dir>)
  2. python3 tools/scripts/validate-evidence.py <dir> — violations
  3. python3 tools/scripts/coverage.py <dir> — phase map + focus
  4. output/status.json — dependencies[] (ages, deadlines, blocks; legacy pending.* values count too) and steps statuses
  5. Last 2-3 entries of output/decisions.md — what was promised last time
  6. CONTEXT.md — is there a target? kill criteria?

Diagnosis order (first match wins the top slot; report up to 3)

  1. Stalled external dependency — an entry in dependencies[] past its deadline (or a legacy pending older than 7 days — legacy support ends in 2.0) → recommend: chase the owner / move the deadline / switch to synthetic (downgrading dependent hypotheses) / consciously skip (status: "skipped"). Name the blocked hypotheses from blocks.
  2. Evidence violations — flagged data_inconsistency held above 0.6, confidence outside type range → recommend reconciling before anything built on those numbers (especially before a gate).
  3. Ready-but-unstarted work — e.g. ≥2 hypotheses confirmed REAL and /solutions not run; solutions scored and /sketch empty; deck ready and gate not rehearsed → recommend the job that converts existing evidence into progress.
  4. Frame gaps that now matter — no target/kill criteria AND a gate is near → recommend /setup-initiative (this is the only moment to push it).
  5. Open hypotheses without a validation path — testing/draft hypotheses that no brief or dependency covers → recommend /brief or /ingest.

Output format

Short. For each recommendation: what → why now → one command/action to start. Reference hypothesis ids and real dates/ages. No generic advice ("keep validating") — every line must be executable today. If everything is genuinely blocked on external people, say exactly that and list who owes what since when.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.