Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add iamcxa/kc-claude-plugins/plugin install kc-dev-flow-2Wrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/iamcxa/kc-claude-plugins/observer)<a href="https://agentmods.dev/skills/iamcxa/kc-claude-plugins/observer"><img src="https://agentmods.dev/badge/skills/iamcxa/kc-claude-plugins/observer/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/iamcxa/kc-claude-plugins/observer"><img src="https://agentmods.dev/badge/skills/iamcxa/kc-claude-plugins/observer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00068 | $0.00991 |
| Opus 5.5 | $0.00027 | $0.00396 |
| Sonnet 5.5 | $0.00014 | $0.00198 |
| Haiku 4.5 | $0.00007 | $0.00099 |
Grade A, and why
observer 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 today.
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 — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Observer
Invocation: kc-dev-flow-2:observer
Package root: ${CLAUDE_PLUGIN_ROOT}
<package> in this skill is this root. If the value above is not an absolute path,
the root is two directories above this SKILL.md.
Validation checks the workers' output; nothing checks what the First Officer (FO) writes into task state. You read those writes as someone who was not there. The caller supplies only these inputs, never its own account of the task:
STATE: the state checkout;TASK: the task slug;RULES: the adopted workflow README.FROMandTO: state commits bounding the window. When not given,FROMis the commit that added the task file andTOis the state checkout'sHEAD.
Read
git -C STATE log --format='%h %ad %s' --date=iso FROM..TO -- <task file> <task dir>and the diff over the same range. Read the task atTOonly around changed hunks. For an archived task give git both path pairs,<TASK>.md <TASK>and_archive/<TASK>.md _archive/<TASK>, and takeFROMfrom the commit that added<TASK>.md; the archive commit alone leaves a window of one commit.python3 <package>/scripts/decision_ledger.py STATE, keeping the rows forTASK: each gate ruling with its question, who recorded it, its grant and anyOverride:.- From RULES collect every sentence whose subject is FO ("FO records", "FO appends", "FO declines", "FO reserves", "FO names", "FO counts", "FO applies") and that applied in this window. Quote each with its section heading.
Check at least: a Captain ruling quotes his words and answers a question put before it;
an approval read as accepting a recommendation has that sentence in its question; a
decision by agent:first-officer carries the grant it relies on; an Override: line
exists where his choice differs from the stated recommendation; a Precedent: names an
ADR and quotes its Words line, or its decision sentence when marked (no Words line),
and is not one of the always-asked kinds; amendments and
Asked: lines follow ## Captain amendments; declines follow step 5 of review findings.
Where RULES list peers under extensions:, also check that no Captain ruling in the task
quotes a peer's words, or a peer's report of the Captain, in place of his own; a ruling he
made in his own words after reading peer data is his. An exchange the file does not record
is outside what state shows.
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.
- today Changed · +2 lines 7512c35ce4e7
- 4d ago Changed · +3 lines aefcef3929ab
- 5d ago First seen · 69 lines · 68 tokens per session scan A 9b98f527e2ce
observer is a skill published in the GitHub repository iamcxa/kc-claude-plugins (3 stars, last pushed today), licensed MIT. It adds 68 tokens to every session and 991 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-10-13.
Other skills, from other repositories
autoreview
Structured code review through the shared OpenClaw agent-skills installation.
omh-code-review
This is an OMH code-review workflow skill, projected for Agent Skills hosts (Claude Code, Codex, Cursor, opencode, OpenClaw, pi).
revdiff-plan
Review the last Codex assistant message (plan, analysis, or proposal) with inline annotations in a TUI overlay. Extracts the most recent response from Codex rollout files and opens it in revdiff for review and annotation. Activates on "revdiff-plan", "review plan with revdiff", "annotate plan", "review last response"…
asplos-writing-style
Use when drafting or revising ASPLOS prose — engineering the first two pages that rapid review actually reads, stating the cross-layer insight as one quotable sentence, structuring per-layer mechanism sections, writing bounded claims, and fitting everything into 11 self-contained pages that include figures and…
explore-unknowns
Guide the user through a quadrant walk that maps the unknowns of a task — open by listing the known knowns, then work through known unknowns, unknown knowns, and unknown unknowns one stage at a time, ending with a complete four-quadrant map in the user's hands. Use when a request is ambiguous or underspecified, the…
re0-merge
Review and land an external contribution the way this suite does: gate it against the thesis, land it with the author's credit intact, complete a new skill rather than merging it raw, then approve, credit, and explain before closing. Use when reviewing a pull request, as any collaborator or maintainer, not only the…