Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/nmahavishnu/planloop/planloopnpx skills add nmahavishnu/planloop --skill planloopgit clone --depth 1 https://github.com/nmahavishnu/planloopWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00128 | $0.02240 |
| Opus 5 | $0.00064 | $0.01120 |
| Sonnet 5 | $0.00026 | $0.00448 |
| Haiku 4.5 | $0.00013 | $0.00224 |
Grade A, and why
planloop scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 126 lines — stays where its author put it; the contents beside it link to each section on GitHub.
planloop
Ask. Plan. Delegate. Verify. Ask again. State lives on disk, not in conversation.
At invocation, check for a learnings file at planloop.learnings.md in the user config directory (~/.claude/ for Claude Code, ~/.codex/ for Codex, ~/.gemini/ for Gemini CLI, and so on). If it exists, its lines tailor how you work and fill gaps. They do not override the core loop: a learning can change how you ask or verify, not whether you plan, get approval, and check. If it does not exist, do not create or mention it.
Step 0: Size the task
Run the loop if the task leaves anything to guess at, if a requirement is unresolved, or if the work is large enough to need slices or to outlive this session. Skip it only when the task is fully specified and small. Skipping means going straight to work with one confirming sentence, no questions, no plan file. The user invoking planloop by name always runs the loop.
Step 1: Ask
One batched round of questions up front. Questions that cannot be answered yet go into plan.md under Open Questions, and each is raised when the work reaches the slice that needs it, not now.
Rules, in order:
- Infer first. Mine the request, the codebase, existing conventions, and prior conversation before asking. Anything inferable becomes a stated assumption, not a question, unless a wrong guess would be costly or hard to reverse, in which case it is a question no matter how inferable.
- Decision relevance test. A question earns a slot only if the answer would change plan.md. If every answer leads to the same plan, do not ask.
- Assumptions as vetoes. Output format:
- "Assumptions I am proceeding with (veto any): A, B, C."
- "Questions I cannot infer: 1, 2." Silence on an assumption is consent.
- Proportional cap. Max 2 questions when only one thing is unresolved. Max 5 when several are, or when the task is large. Never more.
- Group related questions. Fold related concerns into one question; split only when they are genuinely separate.
- One line answers. Every question is closed form with a proposed default: "A or B? I would pick A because X."
- Security. If the task plausibly touches auth, user data, external input, or tool/MCP integrations, ask about it. When the surface is unclear, ask. When it is obviously absent, do not.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 126 lines · 128 tokens per session scan A de93ffb82506
planloop is a skill published in the GitHub repository nmahavishnu/planloop (2 stars, last pushed 1mo ago), licensed MIT. It adds 128 tokens to every session and 2,240 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
apideck-connector-coverage
Check Apideck connector API coverage before building integrations. Use when determining which operations a connector supports, comparing connector capabilities, or diagnosing why an API call fails with a specific connector. Teaches agents to query the Connector API for real-time coverage data.
technical-documentation
Audit, write, and improve developer documentation using Google's Developer Documentation Style Guide and Technical Writing courses. Use this skill for any documentation work, even when the user names no style guide: "audit our docs", "review this README", "write a README", "getting started guide", "how-to or…
ctf-writeup
Generates a single standardized submission-style CTF writeup for competition handoff and organizer review. Use after solving a CTF challenge to document the solution steps, tools used, and lessons learned in a structured format.
checkpoint
Save a structured state snapshot before stopping or handing off. Captures the active plan, recent decisions, file pointers (with line numbers), open questions, and the next 1–3 actions into a checkpoint file under qualityreports/checkpoints/. Optionally proposes [LEARN] entries to add to MEMORY.md. Use when user says…
promote-memory
Review candidate learnings in Claude Code's native auto memory (/.claude/projects/ /memory/, machine-local) and run them through a five-critic council in parallel: generality, staleness, redundancy, evidence, format. Majority vote (3+ of 5) promotes the entry to MEMORY.md. Use when user says "promote memory", "review…
respond-to-eval
Turn student course evaluations (free-text + numeric) into an actionable teaching-improvement plan — the teaching analogue of /respond-to-referees. Clusters comments into themes, separates signal from noise, classifies each theme Keep / Change / Investigate / Out-of-scope, and drafts concrete changes mapped to the…