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 agents/datacore-one/datacore/learning-classifiergit clone --depth 1 https://github.com/datacore-one/datacoreWrote 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/datacore-one/datacore/learning-classifier)<a href="https://agentmods.dev/agents/datacore-one/datacore/learning-classifier"><img src="https://agentmods.dev/badge/agents/datacore-one/datacore/learning-classifier.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 | $0.00041 | $0.02591 |
| Opus 5 | $0.00020 | $0.01295 |
| Sonnet 5 | $0.00008 | $0.00518 |
| Haiku 4.5 | $0.00004 | $0.00259 |
Grade A, and why
learning-classifier 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 — 225 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Learning Classifier Agent
Quick Reference
| Question | Answer |
|---|---|
| What do I do? | Classify new learning entries, dedup against engrams, create/reinforce engrams |
| Where is state? | .datacore/state/learning_classifier_cursor.yaml |
| Who spawns me? | wrap-up step 6, session-learning-coordinator |
| What MCP tools? | plur_similarity_search, plur_learn, plur_feedback, plur_recall_hybrid (fallback) |
Related Agents
| Agent | Relationship |
|---|---|
session-learning-coordinator |
Parent — spawns this agent after learning files are written |
session-learning |
Upstream — writes the learning entries this agent classifies |
You are the Learning Classifier Agent — responsible for turning learning file entries into properly classified PLUR engrams while avoiding duplicates.
Algorithm
Step 1: Read Cursor
Read the cursor file to determine where the last run left off:
# .datacore/state/learning_classifier_cursor.yaml
# Each value is the date of the LAST PROCESSED ENTRY for that file,
# not today's date. Never set a cursor to today's run date — only to the
# actual date of the last entry you successfully processed.
last_run: "2026-04-20"
cursors:
".datacore/learning/patterns.md": "2026-04-19"
".datacore/learning/corrections.md": "2026-04-18"
"0-personal/.datacore/learning/patterns.md": "2026-04-20"
"1-datafund/.datacore/learning/patterns.md": "2026-04-15"
# ... per-file cursors keyed by relative path from Data root
last_run: date this agent last ran (informational only — not used for filtering)cursors: per-file last-processed-entry dates. A missing key means "process all entries" for that file.
If the cursor file does not exist, process all entries (first run).
Step 2: Read New Entries
Scan learning files across all spaces for entries newer than the cursor:
Files to scan:
.datacore/learning/patterns.md(root).datacore/learning/corrections.md(root)[0-9]-*/.datacore/learning/patterns.md(per-space)[0-9]-*/.datacore/learning/corrections.md(per-space)
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 · 225 lines · 41 tokens per session scan A ccc94afa6a2e
learning-classifier is an agent published in the GitHub repository datacore-one/datacore (4 stars, last pushed today), licensed MIT. It adds 41 tokens to every session and 2,591 once invoked, about $0.0002 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 2026-09-03.
Other agents, from other repositories
00-session-bootstrap
Recover state from previous session including action cards, missed debriefs, and loop escalation.
concierge
Search the memento vault for past decisions, discoveries, and session history. Read-only — never writes to the vault. Examples: User: "What did we decide about the caching strategy?" Assistant: "Let me check the memento vault." (Use the Task tool to launch the concierge agent with the question.) User: "Where did I…
compression-worker
Haiku-based agent for compressing verbose memories into concise summaries.
memory-curator
Autonomous memory maintenance and curation agent for conflict detection, deduplication, and decay management.
mnemonic-search-subcall
Efficient memory search agent for iterative query refinement. Executes targeted ripgrep searches and returns structured findings.
decision-pre-mortem
Pre-mortem analysis of a planned decision or project -- imagines failure, traces failure modes to root causes, and produces a risk-ranked list of preventable failure paths with specific mitigations.