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/Goldentrii/AgentRecall-XWrote 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/commands/goldentrii/agentrecall-x/arreflect)<a href="https://agentmods.dev/commands/goldentrii/agentrecall-x/arreflect"><img src="https://agentmods.dev/badge/commands/goldentrii/agentrecall-x/arreflect.svg" alt="Measured on agentmods" 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 | $0.00024 | $0.00984 |
| Opus 5 | $0.00012 | $0.00492 |
| Sonnet 5 | $0.00005 | $0.00197 |
| Haiku 4.5 | $0.00002 | $0.00098 |
Grade A, and why
arreflect 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 8d 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 — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/arreflect — Consolidation & Reflection Loop
Run periodically (every 5-10 sessions, or whenever things feel like they're repeating). Purpose: catch corrections that keep recurring despite already being "known," and turn them into durable rules — with the human approving every rule change.
When to Run
- You've corrected the agent on the same kind of mistake more than once
ar statsshows a growing corrections count with no matching rule update- End of a long project phase, before starting the next one
- The user asks to "reflect," "consolidate what we've learned," or 复盘
SOP
Step 1 — Pull the Current State
ar stats # corrections, feedback, insights, graph edges — health snapshot
ar corrections rejected --stats # survivorship-bias probe: what the capture gate discarded
ar mirror # first-person, citation-backed self-model from real corrections/insights
Read the output. Note which corrections keep showing up, and whether ar mirror reflects a pattern the human hasn't seen named yet.
Step 2 — Confirm Recurrence
For each correction pattern that appears more than once:
does an existing rule (CLAUDE.md / rules/*.md) already cover this?
IF yes AND the violation happened AFTER the rule was written:
the rule exists but didn't take — flag for re-abstraction (Step 4)
IF yes AND the violation happened BEFORE the rule was written:
already covered — no action needed
IF no:
new, unclassified pattern — cluster it (Step 3)
Step 3 — Cluster Unclassified Patterns
Group semantically-similar uncovered corrections into candidate "error classes" (e.g. "forgets to gate destructive git ops," "assumes global binaries exist"). For each cluster of 2 or more items, draft:
- a short class name
- a one-sentence description
- the keywords/triggers that identify it
Present clusters to the human. Don't invent a class for a single one-off correction.
Step 4 — Run Consolidation and Draft Re-Abstractions
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.
- 8d ago First seen · 98 lines · 24 tokens per session scan A 86edfa39618a
arreflect is a command published in the GitHub repository Goldentrii/AgentRecall-X (371 stars, last pushed 7d ago), licensed MIT. It adds 24 tokens to every session and 984 once invoked, about $0.0001 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 2026-08-30.
Other commands, from other repositories
evolve-lite-provenance
Analyze saved trajectories and recall audit events offline to record whether recalled guidelines influenced completed sessions.
evolve-lite-retention
Apply data-retention rules to the local evolve store — flag or delete stale and unused memories and expired sessions (dry-run by default).
evolve-lite-adapt-memory
Mirror a just-saved native memory into the shared evolve store so it becomes shareable and auditable.
artel-remember
Write a fact, decision, or gotcha to Artel shared memory.
dream
Run a judgment session over the Pseudolife memory bank — triage the review queues; extract facts only where no extractor can.
artel-recall
Search Artel shared memory for what the fleet already knows.