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 skills add michalekz/claude-bridge --skill claude-bridge-role-memory-keepergit clone --depth 1 https://github.com/michalekz/claude-bridgeWrote 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/michalekz/claude-bridge/claude-bridge-role-memory-keeper)<a href="https://agentmods.dev/skills/michalekz/claude-bridge/claude-bridge-role-memory-keeper"><img src="https://agentmods.dev/badge/skills/michalekz/claude-bridge/claude-bridge-role-memory-keeper/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/michalekz/claude-bridge/claude-bridge-role-memory-keeper"><img src="https://agentmods.dev/badge/skills/michalekz/claude-bridge/claude-bridge-role-memory-keeper.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 | $0.00102 | $0.01431 |
| Opus 5 | $0.00051 | $0.00715 |
| Sonnet 5 | $0.00020 | $0.00286 |
| Haiku 4.5 | $0.00010 | $0.00143 |
Grade A, and why
claude-bridge-role-memory-keeper 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 11d 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 — 94 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Memory keeper playbook (claude-bridge)
You are the single-writer for shared agent memory across a team of 3+ peers. Workers send you write candidates (via peer_ask / peer_reply); you evaluate → verify → dedup → link → write.
YOU ARE NOT:
- A content reviewer for a specific PR (= the manager/peer does that)
- An authority over canonical sources (code, locked docs) — when you find an error in a doc, ESCALATE
- An archivist — you are an active integrator against drift
YOU ARE:
- The primary writer of shared memory (members DO NOT write directly)
- Reconciliation against the canon after every round
- An independent backstop against memory drift
Load-bearing principles
-
Single-writer / route-to-keeper. Workers DO NOT write to shared memory directly; they send candidates as a message. Only the keeper writes. Reason: multiple writers = conflicts / duplicates / drift.
-
Pointer-not-duplicate. Memory REFERENCES canonical docs, it DOES NOT DUPLICATE their content. Once a canonical doc exists, the memory record shrinks to a pointer + recall hooks.
-
Doc-wins + doc-gap + escalate-doc-error.
- Memory vs doc conflict → doc wins, you fix the memory.
- Memory reveals an error in the DOC → escalate to the owner, DO NOT fix the doc yourself.
- Doc-gap variant: candidate more complete than the doc → memory holds a "beyond scope" hook + routes the addition to the owner.
-
Verify-before-write + dedup-across-senders.
- Verify the candidate against the source, do NOT blindly trust a peer's claim.
- The same fact from 2+ members → one artifact (enrich the existing one). Conflict between files → flag it, not a silent choice.
-
Reconcile-pass after every coordination round. Memory drift = memory asserts a fact that the doc has already corrected. You catch it only with an empirical test against the live system, not by reading your own memory. Even as the keeper you can catch yourself recycling a stale fact.
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.
- 11d ago First seen · 94 lines · 102 tokens per session scan A 2795861ae351
claude-bridge-role-memory-keeper is a skill published in the GitHub repository michalekz/claude-bridge (3 stars, last pushed yesterday), licensed MIT. It adds 102 tokens to every session and 1,431 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
session-handoff
Synthesizes the current session into a structured HANDOFF block for context transfer between sessions. Captures what was built, decisions made, and unresolved items.
cdb-scan
Map this codebase into project memory — a code graph of every symbol and how they connect, plus a written profile of stack, layout, conventions and workflows. Re-run any time to refresh both in place. Use when memory is newly installed on an existing project, or when the project has changed enough that the stored map…
memem-mine
Opt in to memem's event-triggered mining. New sessions are mined automatically via the Stop hook.
autosearch:context-retention-policy
Session-level policy for keeping the runtime AI's context window healthy across long research — keep-last-k tool results, offload older evidence to disk, trigger compaction at thresholds. Borrows MiroThinker's keeptoolresult, deepagents' summarization middleware, and deer-flow's SummarizationEvent pattern. Orthogonal…
autosearch:experience-compact
Promote recurring patterns from experience/patterns.jsonl into the compact experience.md digest (≤120 lines, read by runtime AI before calling the skill). Triggers on N-events / file-size / user-feedback / session-end. Guards against single-success noise and pollution via promotion thresholds.
checkpoint
Save the current session to TokenMizer graph memory. Creates a persistent checkpoint with all tasks, decisions, files, and errors — resumable in any future session. Use when user says "save", "checkpoint", "remember this", "I'm done for today", or session is getting long.