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 agentmods add commands/adilkalam/orca/reflectgit clone --depth 1 https://github.com/adilkalam/orcaWhat 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 | $0.00023 | $0.00654 |
| Opus 5 | $0.00012 | $0.00327 |
| Sonnet 5 | $0.00005 | $0.00131 |
| Haiku 4.5 | $0.00002 | $0.00065 |
Grade A, and why
reflect 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 yesterday.
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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/reflect -- Institutional Learning (Learned Rules curation)
/reflect is the manager for the Learned Rules ledger in CLAUDE.md (the
## Learned Rules (via /reflect) section). It is a thin wrapper over two scripts that
already do the work:
~/.claude/scripts/reflect-analyze.py-- mines learning signals (corrections, standing instructions, negative feedback) from session history.~/.claude/scripts/reflect-apply.py-- reads/writes the Learned Rules section (add/archive/remove/list).
Rules are NEVER applied automatically. /reflect proposes; the owner confirms each
addition. This is the one command allowed to edit the Learned Rules section.
Subcommands
/reflect status (default)
Show the current Learned Rules ledger.
python3 ~/.claude/scripts/reflect-apply.py list
/reflect --source recording|jsonl [--days N]
Mine learning signals from session history. Data source auto tries .orca/recording.db
first, then falls back to JSONL transcripts; recording / jsonl force one.
python3 ~/.claude/scripts/reflect-analyze.py --days ${N:-30} --source ${SOURCE:-auto} --format summary
Present the ranked signals to the owner. Do NOT write anything in this mode.
/reflect learn [--days N]
The full loop, with a mandatory human gate:
- Run
reflect-analyze.py(as above) to surface candidate rules. - For each candidate the owner approves (
AskUserQuestion), apply it:
Soft/preference-level learnings go topython3 ~/.claude/scripts/reflect-apply.py add --rule "<approved rule text>" --target claude_md--target workshopinstead. - Show the resulting diff of the Learned Rules section and stop. Never batch-apply without per-rule confirmation.
Curation discipline
- The ledger is bounded: prefer editing/merging an existing rule over adding a near-duplicate.
- Archive (don't delete) rules that are superseded:
reflect-apply.py archive --rule-id <id> --reason "<why>". - Rule text must be a single imperative sentence a future session can act on.
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.
- yesterday First seen · 61 lines · 23 tokens per session scan A 3c606b4a12d2
reflect is a command published in the GitHub repository adilkalam/orca (2 stars, last pushed 1mo ago), licensed MIT. It adds 23 tokens to every session and 654 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 2026-08-31.
Other commands, from other repositories
learn
Force claude-smart to extract learnings from this session now.
component
Scaffold a new React component grounded in the paper-mono primitives. Requires explicit kind or a nearest-existing-component match. No empty divs, no speculative scaffolding.
memory-store
Store an insight, decision, or pattern to memory.
review
Cold re-quiz on code that already shipped — your own session commits, not the change in front of you.
no-vibe
Enter no-vibe mode in OpenCode (tutor mode, no direct project file writes).
teach-me-testing
Teach testing progressively through structured sessions. Use when user says ""lets learn testing"" or ""I want to study test practices"".