Blocker escalation report
Skill KingSharkG/ai-agents-workflow/skills/blocker-escalation-report
Produce a structured Blocker Escalation Report when the workflow cannot progress — used after review cycle 3 is exhausted, a missing-context blocker cannot be resolved, or an agent reaches an unresolvable decision point.From its SKILL.md
npx -y skills add KingSharkG/ai-agents-workflow --skill blocker-escalation-reportAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
4.1 KB, as published. Nobody here has run it
Blocker Escalation Report
Use this skill to produce a Blocker Escalation Report. This is the structured exit path when the workflow cannot progress and requires human input or re-routing.
When to Use
- Review cycle count has reached the complexity-tied cap and issues remain unresolved
- An executor or lead agent cannot proceed without information not present in the artifact chain
- An ambiguous requirement cannot be resolved by the responsible Lead's domain-validation pass
- An external dependency (third-party API, undelivered upstream) blocks implementation
- Executor or Lead detects the Delivery Plan itself is defective (route_to: delivery-pm)
- The chief-orchestrator must escalate to the user rather than silently stalling
- The required
section:tep,section:spec, orsection:implementationplaceholder is absent fromai-work.md
Output Target
Append to <!-- section:escalation-N --> in the subtask's ai-work.md. The orchestrator assigns N (incrementing per escalation event within the subtask) and writes the placeholder before this skill runs. Also write the diagnostics footer to <subtask_id>/summary.md per ${CLAUDE_PLUGIN_ROOT}/skills/orchestrator-telemetry/references/artifact-footer-protocol.md (role: the blocked agent's role; telemetry-line status: escalated).
Output Format
Append inside <!-- section:escalation-N -->:
<!-- section:blocker-metadata -->
## Metadata
- **task_id**: <from the originating task>
- **subtask_id**: <from the Delivery Plan, if applicable>
- **blocked_agent**: <name of the agent that hit the blocker>
- **escalated_by**: <name of the agent producing this report>
- **timestamp**: <ISO 8601 UTC>
- **cycle_count**: <current review cycle number, if applicable>
<!-- /section:blocker-metadata -->
<!-- section:blocker-type -->
## Blocker Type
<!-- Select one: -->
- [ ] missing-context — required information is absent from all artifacts
- [ ] ambiguous-requirement — Task Packet or PROJECT_CONFIG baseline is unclear and inference would be unsafe
- [ ] external-dependency — blocked on an undelivered or third-party resource
- [ ] cycles-exhausted — complexity-tied review cycle cap reached without resolution
- [ ] plan-defective — Delivery Plan scope / dependencies / subtask boundaries are wrong (route_to: delivery-pm)
- [ ] missing-skeleton-section — required section placeholder absent from ai-work.md
<!-- /section:blocker-type -->
<!-- section:blocker-what-is-blocked -->
## What Is Blocked
<!-- One sentence: the specific action that cannot proceed. -->
<!-- /section:blocker-what-is-blocked -->
<!-- section:blocker-what-was-tried -->
## What Was Tried
<!-- Bullet list: what the blocked agent attempted and why it was insufficient. -->
<!-- /section:blocker-what-was-tried -->
<!-- section:blocker-required-input -->
## Required Input
<!-- Precise question(s) or decision(s) needed to unblock. -->
<!-- /section:blocker-required-input -->
<!-- section:blocker-suggested-rerouting -->
## Suggested Re-routing
<!-- Set `route_to:` — one of: lead | delivery-pm | user. -->
<!-- /section:blocker-suggested-rerouting -->
Then write the diagnostics footer (role: blocked agent's role; telemetry-line status: escalated) per the artifact-footer protocol referenced above.
Rules
- A Blocker Escalation Report is a terminal artifact for its subtask. Do not attempt further implementation while it is open.
- The chief-orchestrator must surface the
required_inputto the user immediately after receiving this artifact. - If re-routing to Delivery PM when the plan is defective, attach the escalation section content as context — do not summarize it in prose.
- After resolution, the orchestrator updates the TEP section or triggers a Lead re-run; do not resume from the stalled implementation section.
- The orchestrator's Artifact Gate still applies — section must be non-empty after the agent returns.