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 ychampion/cskill-agents --skill mark-after-filter-memory-deduplicationgit clone --depth 1 https://github.com/ychampion/cskill-agentsWrote 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/ychampion/cskill-agents/mark-after-filter-memory-deduplication)<a href="https://agentmods.dev/skills/ychampion/cskill-agents/mark-after-filter-memory-deduplication"><img src="https://agentmods.dev/badge/skills/ychampion/cskill-agents/mark-after-filter-memory-deduplication/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/ychampion/cskill-agents/mark-after-filter-memory-deduplication"><img src="https://agentmods.dev/badge/skills/ychampion/cskill-agents/mark-after-filter-memory-deduplication.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 | $0.00035 | $0.00455 |
| Opus 5.5 | $0.00014 | $0.00182 |
| Sonnet 5.5 | $0.00007 | $0.00091 |
| Haiku 4.5 | $0.00003 | $0.00046 |
Grade A, and why
mark-after-filter-memory-deduplication 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 1mo 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 — 28 lines — stays where its author put it; the contents beside it link to each section on GitHub.
SKILL: Mark-After-Filter Memory Deduplication
Domain: context-management
Trigger: Apply when asynchronously prefetched memories or file attachments must be deduplicated against a cumulative read-state cache that they will also update after surviving the filter.
Source Pattern: Distilled from reviewed context, compaction, and memory-governance patterns.
Core Method
When prefetch results need to consult a cumulative "already seen" index, treat that index as read-only during selection. First filter the candidate attachments against the existing state, then write only the survivors into the shared cache. This breaks the self-referential cycle where a prefetch marks everything as seen before dedupe runs, causing the filter to delete the very data it just found. Keep the dedupe state cumulative across the whole turn so later iterations and future turns still suppress genuine repeats.
Key Rules
- Read from the shared dedupe state during filtering, but do not mutate it until the final survivor set is known.
- Update the shared cache only for attachments that actually survive the filter and remain eligible for surfacing.
- Use a cumulative state store that includes earlier loop iterations and prior surfaced memories, not just the current batch.
- Keep the mutation step adjacent to the filter so future maintainers cannot reintroduce self-poisoning writes inside the prefetch path.
Example Application
If a background document selector prefetches five candidate help files while the main agent is still running tools, first drop any files already present in the session's read-state map. Only after that should you record the surviving files in the map, ensuring they surface once now and stay suppressed on later retries.
Anti-Patterns (What NOT to do)
- Do not mark prefetched files as read before duplicate filtering; the filter will see its own writes and erase the whole batch.
- Do not dedupe only against the current iteration's attachments; you need cumulative session state to avoid resurfacing files already read by tools or earlier turns.
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.
- 1mo ago First seen · 28 lines · 35 tokens per session scan A b940ff593096
mark-after-filter-memory-deduplication is a skill published in the GitHub repository ychampion/cskill-agents (36 stars, last pushed 6mo ago), licensed MIT. It adds 35 tokens to every session and 455 once invoked, about $0.0001 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 2026-09-03.
Other skills, from other repositories
grape
Use Grape MCP for Codex context continuity in coding repositories. Use when a task needs repeated-turn context, omitted context restore, stale-context checks, invalidation checks, or safe continuity across branch and dirty-worktree changes.
brain-page
Operating manual for reading and writing a project's brain — every read and write goes through the bundled zero-dependency brain CLI; never hand-edit brain files. Read it before creating or modifying any page or root page.
brain-setup
Bootstrap the Open Project Brain Standard into the current project — prefer brain init (ensure BRAIN.md, scaffold empty brain brainRoot-aware, default-wire CLAUDE.md + AGENTS.md). Optionally install a pre-commit hook and a Claude Code or Codex SessionStart hook.
brain-bootstrap
Seed a freshly-scaffolded brain with real project knowledge — on an existing (brownfield) project read the code, docs, and git log, then draft only the root pages that have meaningful content and capture key historical decisions; on a near-empty (greenfield) project interview the user. Every write goes through the…
brain-ingest
The process for digesting a conversation, document, or research result, classifying it, and writing it down as brain content (a root-page update or a new/updated page) through the brain CLI.
learn
Diagnose and fix agent behavioral surfaces when the user corrects a mistake — connects to Claude native memory.