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/iamk77/skill/groundworknpx skills add IamK77/Skill --skill groundworkgit clone --depth 1 https://github.com/IamK77/SkillWhat 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 | $0.00130 | $0.02830 |
| Opus 5 | $0.00065 | $0.01415 |
| Sonnet 5 | $0.00026 | $0.00566 |
| Haiku 4.5 | $0.00013 | $0.00283 |
Grade A, and why
groundwork 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 — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.
groundwork
!checklist init ${CLAUDE_SKILL_DIR} --force
Building the wrong thing well is the most expensive mistake in software, and an agent's strongest instinct — implement the literal ask — walks straight into it. This skill is the discipline that runs before code: discover the real need, the people it touches, the unspoken constraints, and the explicit edges of scope. Requirements work is a collaboration — you supply systematic rigor, the user supplies domain truth and the stakes.
This skill walks an explicit five-stage flight plan and will not fly past a GATE until it is cleared.
Discipline: finish every GATE before entering the next stage. GATEs are hard — never skip, batch past, or self-certify a stage you have not actually done. The checklist tool enforces the order; let it. Commands address stages by name, so you never have to track a number.
The agent's special handicap: you cannot walk the floor, run a workshop, or interview the silent stakeholder. So you must enlist the user as proxy for everyone the requirement touches, and you must read what can be read — the existing code, docs, tickets, competitors. Read references/agent-blind-spots.md first; it names the failure modes this skill exists to stop.
Speak the user's language, or the proxy breaks. The user can only stand in for the stakeholders if they actually follow you. Read their domain fluency from how they talk in STAGE 0, and match it — when a technical term or acronym first appears (SSRF, BYOK, p95, idempotent), gloss it in half a sentence. Skipping that is not brevity; it buys a confirmation that confirms nothing — a user nodding at words they did not parse, which makes STAGE 3's "what I understood = what you want" check pass falsely.
When there is no human in the loop (design-only). Sometimes no user is reachable for the whole run — you are scoping an OSS feature from an RFC, or designing against a written spec alone. The skill still applies, in a shifted mode: the invariant never guess silently on a high-blast-radius unknown still binds — its satisfaction shifts from get a yes to park the unknown as a named open question with an owner-to-ask and its blast-radius-if-wrong. Re-point the validation oracle from user intent to the best available written authority (the RFC, an enumerated reviewer comment, a standard like RFC 7519, the existing code) — see the WRITTEN-PROXY rule in references/decision-tree.md. STAGE 3 (Validate) then degrades to replay every Must against its written source + walk the full review checklist + park the unconfirmable; STAGE 4 (Manage) produces a proposed governance contract for whoever inherits the work, not a live authority gate. This is a weaker claim than user-confirmed understanding — label it as such in the handoff. The conservative-default FALLBACKs scattered through the references are this one mode; read them as one.
If $ARGUMENTS names a target to scope, scope that. If the ask is already a fully-pinned, low-risk change (a one-line fix, a rename), this skill is overkill — say so and proceed without it. When in doubt, open the decision tree.
What ships with it
9 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 142 lines · 130 tokens per session scan A 6bb7ab6c0921
groundwork is a skill published in the GitHub repository IamK77/Skill (2 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 130 tokens to every session and 2,830 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
claude-md-improver
Audit and improve CLAUDE.md files in repositories. Use when user asks to check, audit, update, improve, or fix CLAUDE.md files. Scans for all CLAUDE.md files, evaluates quality against templates, outputs quality report, then makes targeted updates. Also use when the user mentions "CLAUDE.md maintenance" or "project…
agent-platform-rag-engine-management
Manage and query Agent Platform RAG Engine Corpora and retrieve grounded contexts using the Google GenAI SDK. Use when listing RAG corpora or files, inspecting a corpus, retrieving contexts, or generating content grounded in a RAG corpus. Do not use for standard database queries (use SQL/Spanner skills), Google…
agent-platform-model-registry
Agent Platform Model Registry Management. Use when you need to upload, list, describe, update, or delete machine learning models (and their versions) in the Agent Platform Model Registry. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform models.
offensive-c2-frameworks
Command and Control framework deployment, configuration, and operational tradecraft for red team engagements. Covers Cobalt Strike (malleable C2 profiles, Beacon types HTTP/HTTPS/DNS/SMB, Beacon Object Files for in-memory execution, sleep and jitter tuning, named pipe pivoting), Sliver (implant generation across…
twitter-reader
Read Twitter/X for financial research using opencli (read-only). Use this skill whenever the user wants to read their Twitter feed, search for financial tweets, view bookmarks, look up user profiles, or gather market sentiment from Twitter/X. Triggers include: "check my feed", "search Twitter for", "show my…
quick-recap
Use when adding or following the red/yellow/green final status block convention for agent responses, especially by installing managed AGENTS.md or CLAUDE.md instructions.