mikekelly/recovering-subagents
plugins/promode/skills/recovering-subagents/SKILL.md
Inspect a failed or stalled subagent's transcript compactly to recover. Use when a background subagent failed, stalled, or its <result> summary isn't enough to decide what to do next, and you need to look inside its run without loading the whole transcript into your context.
$ install --global
skillsauthnpx skillsauth add mikekelly/promode recovering-subagentsInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 7, 2026, 6:06 AM201.2s3 files scanned
SKILL.md
- name:
- recovering-subagents
- description:
- Inspect a failed or stalled subagent's transcript compactly to recover. Use when a background subagent failed, stalled, or its <result> summary isn't enough to decide what to do next, and you need to look inside its run without loading the whole transcript into your context.
When a subagent fails or stalls and its <result> summary isn't enough to act on, you need to look inside its run — but the transcript (the .output file whose path arrives in the <task-notification>) is large JSONL, and reading it whole overflows your context. This skill exists to honour the promode principle that context is precious: it lets you inspect the run one step at a time instead of paying the full transcript's context cost.
Use the bundled inspector instead. It takes the absolute output-file path (copy it from the notification) and prints one step at a time, compactly, so you walk back from the latest step and expand only what you need. Run it from this skill's directory (the base directory is shown when this skill loads):
scripts/inspect-agent.sh <output-file> # the latest step (the "tip")
scripts/inspect-agent.sh <output-file> <step> # that step, compact
scripts/inspect-agent.sh <output-file> <step> full # that step, full JSON
Steps are 1-indexed transcript lines, chronological; the tip is the latest. Every call prints the current tip, so if a running agent has advanced past where you're walking, you'll see it.
Backtrack from the tip
- Start at the tip to see where the agent ended up — the failure, or its current position.
- Walk backwards one step at a time (
<step>= tip−1, tip−2, …) until you find the first step that went wrong: the decision or tool call that made the failure inevitable. full-expand only the one or two steps you actually need the detail of.- Decide the recovery — re-dispatch with a tighter scope, hand it to
promode:debugger, fix the prompt, etc.
Never read the raw .output file directly — it's the full transcript and will overflow your context. This inspector is the context-safe way in. (Requires jq.)
Related Skills
mikekelly/design-system-lookbook
development
VerifiedTrustedCommunity
Establish a design source-of-truth (a DESIGN.md-style two-layer doc of tokens + rationale), build a lookbook that renders it, and wire a live-refresh preview server so visual work gets a fast edit→see feedback loop. Use when setting up or restructuring a design system / design tokens, creating a DESIGN.md or design source-of-truth, building a lookbook, or wanting live preview / live reload of design or marketing artifacts — landing pages, decks, one-pagers, marketing material previews. The visual analogue of the operator-seam test loop: it crystallises taste into determinism. Defers aesthetic taste (typography/color/motion choices) to the frontend-design skill.
16SKILL.mdUpdated Jun 8, 2026
mikekelly/reinforce-design-constraints
development
VerifiedTrustedCommunity
Walk the knowledge graph (rooted at CLAUDE.md) and ensure every crucial design constraint — invariants, prohibitions, required patterns, load-bearing decisions — is stated inline in the nearest loaded CLAUDE.md orientation that governs the affected area, with a link to its full rationale for expanded discovery. Use when constraints are buried in ADRs, knowledge docs, code comments, or tribal knowledge; when an agent violated a rule it had no way to know; or when asked to surface, hoist, lift, strengthen, or reinforce design constraints, or make them discoverable to agents.
16SKILL.mdUpdated Jun 7, 2026
mikekelly/promode-audit
development
VerifiedTrustedCommunity
Audit how well a repository's codebase and practices align with the promode methodology, then produce a prioritised, actionable improvement plan. Fans out parallel assessors (one per dimension) and synthesises their findings. Use when the user wants to assess promode alignment/fit, audit a repo against the methodology, or get a plan to bring a codebase in line with promode. Also flags stale per-project install leftovers — promode ships its own SessionStart hook, so nothing should be copied into a project.
16SKILL.mdUpdated Jun 7, 2026
mikekelly/handoff
testing
VerifiedTrustedCommunity
Write a handoff document so a fresh agent can continue this work after the conversation ends. Invoke when the user is about to /clear or /compact, when context is filling up, or when the user asks to hand off, checkpoint, or pause work for a later session. Argument (optional): what the next session will focus on.
16SKILL.mdUpdated Jun 7, 2026