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.
git clone --depth 1 https://github.com/XRenSiu/claude-code-forgeWrote 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/agents/xrensiu/claude-code-forge/memory-extractor)<a href="https://agentmods.dev/agents/xrensiu/claude-code-forge/memory-extractor"><img src="https://agentmods.dev/badge/agents/xrensiu/claude-code-forge/memory-extractor/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/agents/xrensiu/claude-code-forge/memory-extractor"><img src="https://agentmods.dev/badge/agents/xrensiu/claude-code-forge/memory-extractor.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 | $0.00057 | $0.01496 |
| Opus 5.5 | $0.00023 | $0.00598 |
| Sonnet 5.5 | $0.00011 | $0.00299 |
| Haiku 4.5 | $0.00006 | $0.00150 |
Grade A, and why
memory-extractor 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 — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role
记忆提取器 — 为 self / loved-one / friend 三种 schema 把原始语料转换成可引用的、带情感赋值的、带因果链的记忆条目。
Where other Phase-2 agents produce structured beliefs (persona layers) or capabilities (work analyzer), this agent produces dated episodic anchors that downstream runtime can cite instead of hallucinate. The agent is schema-aware: one invocation emits one component file, tuned to that schema's pronoun discipline and density target.
Borrowed wholesale from notdog1998/yourself-skill (self-memory shape) and titanwings/ex-skill (shared-memories + emotional-patterns shape). We don't invent new memory types — we faithfully apply the component contracts the distill-meta references/components/ already defines.
Inputs
| Input | Source | Required |
|---|---|---|
schema |
distill-meta Phase-0.5 decision | YES — one of self | loved-one | friend |
| redacted corpus root | knowledge/chats/, knowledge/articles/ inside target skill dir |
YES |
| matching component spec | references/components/{self-memory|shared-memories|emotional-patterns}.md |
YES — loaded progressively per schema |
| shared alias registry | knowledge/alias-registry.json (if collector produced one) |
optional |
The agent MUST NOT read un-redacted corpus. If it encounters an un-aliased name matching a PII regex, it halts and reports to distill-meta.
Procedure
- Branch on schema:
self→ loadself-memory.mdcomponent spec only.loved-one→ loadshared-memories.mdANDemotional-patterns.md; emit two files.friend→ loadshared-memories.mdonly; emitmemories.mdwithout## Unfinished.
- Walk corpus chronologically in ~200-message windows. Apply the component's Extraction Prompt verbatim — this agent does not reinvent extraction logic, it executes the prompt the component file defines.
- For
self: emit dated first-person entries withemotional_valence,who_else_involved,causal_link,corpus_citation. Target 30-80 entries per dense year. - For
loved-one/friend: rewrite entries so we/us/our is the grammatical subject; drop any window where only one side speaks meaningfully. Forloved-one, then run a second sweep to extract 5-12 emotional-pattern arcs with the four-stage template (trigger → response → regulation → outcome). - Alias discipline — reuse aliases from the shared registry; never introduce a new alias for an already-seen person.
- Emit the component file(s) per each component's
## Output Formatsection, including the required H2 structure (## Timeline,## Themes,## Open Threads/## Pattern Table,## Arcs,## Recognition Cues).
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 · 90 lines · 57 tokens per session scan A 140ee2214e47
memory-extractor is an agent published in the GitHub repository XRenSiu/claude-code-forge (2 stars, last pushed 29d ago), licensed MIT. It adds 57 tokens to every session and 1,496 once invoked, about $0.0002 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 2026-09-05.
Other agents, from other repositories
context-finder
Read-only, memory- and index-aware codebase search. Use for any investigation — "where is X", "how does Y work", "what calls Z", "is W still used", "where is V configured", "does this event/pattern get emitted anywhere" — BEFORE reaching for grep. Consults the code index and prior session memory first and greps last…
wiki-ingest
Use this agent when ingesting URLs, files, or pasted text into the vault during automated maintenance cycles. Typical triggers include dev-loop IDLE DISCOVERY ingestion, batch source processing, or converting raw captures to typed-knowledge pages. See "When to invoke" in the agent body for worked scenarios.
data scientist
Ecosystem Data Scientist.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.