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/s-annam/okfmem/implement-batchnpx skills add s-annam/okfmem --skill implement-batchgit clone --depth 1 https://github.com/s-annam/okfmemWhat 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 | $0.00092 | $0.07370 |
| Opus 5 | $0.00046 | $0.03685 |
| Sonnet 5 | $0.00018 | $0.01474 |
| Haiku 4.5 | $0.00009 | $0.00737 |
Grade A, and why
implement-batch 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 — 449 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Implement Batch
Orchestrate implementing a set of GitHub issues — the sub-issues of an epic,
or an explicit list — onto one branch, accumulating their changes into a
single commit, then hardening the result with an adversarial review pass
and opening one PR via /open-pr.
This is a GitHub-only, self-contained batch driver for s-annam/okfmem. It
has no Linear code and no dependency on a global /implement-issue — the
per-issue implementation contract is embedded here, in the subagent spawn prompt,
so this skill works for anyone who clones okfmem (contributors included), not
just a maintainer whose ~/tools/skills/ has the global skills.
Why a subagent per issue. The orchestrator can't
/compactmid-run, and a multi-issue run would blow the main context. Running each issue inside its own subagent is the context-isolation mechanism: the subagent's heavy explore/edit context stays down there; only a tight structured summary returns. The orchestrator stays lean across the whole sequence.
Repo facts (okfmem)
- Repo:
s-annam/okfmem, a public repo.mainis protected — every change merges through a PR that needs a greenverifycheck, linear history, and 1 approving review from a Code Owner (.github/CODEOWNERS). Direct commits/pushes tomainare blocked. So this skill never commits onmain— it works on one feature branch and finalizes through/open-pr. (While the repo is solo, the owner can't self-approve, so they merge via admin bypass onceverifyis green; that bypass ends when a co-maintainer joins.) - Public-repo leak rule is non-negotiable — this replaces the usual
fixture-PII rule. Never let a private string into a tracked file: a real home
path (
/Users/<name>), aclaude.ai/code/session_…URL, aClaude-Session:trailer, a personal email, or private store/session contents. Use the documented placeholders (~/okfmem-store,$OKFMEM_STORE,/Users/you). The gate ispython3 scripts/check-leaks.py— it scans tracked-file content and exits non-zero namingfile:line. It runs first in CI (hard fail) and again in/open-pr's preflight (Step 3.5); flag any new private-looking prose the moment a subagent reports it, since the gate cannot judge whether prose is otherwise private. - Gates:
python3 scripts/check-leaks.py(hard) ·python3 -m pytest tests/·ruff check .(advisory) ·ruff format --check .(advisory). CI'sverifyruns the whole mirror across an OS matrix. Per issue, run only the fast/affected checks — the leak gate + the affectedpytestfiles +ruff checkon the touched files. The full matrixverifyis the PR gate (CI), not a per-issue gate. There is no typecheck / build / fallow step — this is pure-Python stdlib. - Confirmation discipline (okfmem-specific, three-rung ladder). okfmem's own
scripts must never silently mutate outward or user-owned state, and never
destroy without a typed confirmation. If an issue's change adds/edits a
state-changing op, the implementing subagent MUST place it on the right rung
(read-only/additive → no friction; outward/user-config mutating →
[y/N]+ skippable-non-interactively with a printed manual command; destructive → typed confirmation, never a bare-Force). SeeCLAUDE.md→ "Confirmation discipline". Treat a violation as a blocking review finding. - Reviewer agent:
ecc:python-reviewer(pure-Python repo), falling back toecc:code-reviewer. These are maintainer-global (~/.claude/agents/), not in this repo — a fresh clone won't have them. On the default (review-on) path, if noecc:*reviewer resolves, fall back to ageneral-purposesubagent driving the built-in/code-reviewskill; a fresh clone that wants to skip review entirely passes--no-review. - Cross-platform. okfmem ships
.ps1/.cmdwrappers and Windows-aware path code. If an issue touches path handling, encoding, or a shipped.ps1, the subagent must respect the codebase's Windows invariants (pure-ASCII.ps1, path decisions by shape notisdir, expected path-assertion values from the code's own normalizer). Treat a hand-typed POSIX literal in a path assertion as a blocking finding — it fails Windows-only CI.
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 · 449 lines · 92 tokens per session scan A 4fa08aee0cb1
implement-batch is a skill published in the GitHub repository s-annam/okfmem (5 stars, last pushed 19d ago), licensed Apache-2.0. It adds 92 tokens to every session and 7,370 once invoked, about $0.0005 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
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
agent-host-chat-contributions
Build and review cross-cutting agent-host chat behavior through lifecycle contributions. Use when adding turn lifecycle side effects, prompt or context injection, restored-history transformation, protocol-action observation, or when reviewing changes that add code to AgentSideEffects or AgentService.
auto-perf-optimize
Run agent-driven VS Code performance or memory investigations. Use when asked to launch Code OSS, automate a VS Code scenario, run the Chat memory smoke runner, capture renderer heap snapshots, take workflow screenshots, compare run summaries, or drive a repeatable scenario before heap-snapshot analysis.