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 ShaheerKhawaja/ProductionOS --skill continuous-learninggit clone --depth 1 https://github.com/ShaheerKhawaja/ProductionOSWrote 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/shaheerkhawaja/productionos/continuous-learning)<a href="https://agentmods.dev/skills/shaheerkhawaja/productionos/continuous-learning"><img src="https://agentmods.dev/badge/skills/shaheerkhawaja/productionos/continuous-learning.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 | $0.00036 | $0.00749 |
| Opus 5 | $0.00018 | $0.00375 |
| Sonnet 5 | $0.00007 | $0.00150 |
| Haiku 4.5 | $0.00004 | $0.00075 |
Grade A, and why
continuous-learning 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 7d 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 — 100 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ProductionOS Continuous Learning
Architecture
SESSION ACTIVITY (via PostToolUse hooks)
↓
Observation Log (~/.productionos/analytics/skill-usage.jsonl)
↓
Pattern Extractor (runs at session end via Stop hook)
↓
Instinct Creation (~/.productionos/instincts/project/{hash}/)
↓
Confidence Scoring (0.0 - 1.0, based on repetition + outcome)
↓
Auto-Promotion (confidence > 0.8 → global instinct)
↓
Global Instincts (~/.productionos/instincts/global/)
Instinct Format
Each instinct is a single markdown file:
---
name: {pattern-name}
confidence: 0.65
observations: 3
first_seen: 2026-03-20
last_seen: 2026-03-20
scope: project
promoted_from: null
---
## Pattern
{What was observed}
## Evidence
- Session {date}: {observation 1}
- Session {date}: {observation 2}
## Application
{When to apply this pattern in future sessions}
Confidence Scoring
| Confidence | Meaning | Action |
|---|---|---|
| 0.0 - 0.3 | Weak signal | Store but don't act |
| 0.3 - 0.6 | Emerging pattern | Suggest when relevant |
| 0.6 - 0.8 | Strong pattern | Apply proactively |
| 0.8 - 1.0 | Proven pattern | Auto-promote to global |
Observation Types
- Code Pattern — Repeated code structure across sessions
- Tool Usage — Preferred tool sequences (e.g., always Grep before Edit)
- Error Recovery — How errors were resolved (reusable fix patterns)
- Architecture Decision — Repeated structural choices
- Review Finding — Common issues found in code reviews
Stop Hook: Pattern Extraction
At session end, the Stop hook:
- Reads
~/.productionos/analytics/skill-usage.jsonlfor this session - Groups events by type (edit, bash, security_edit, etc.)
- Identifies patterns (repeated file types, common commands, recurring errors)
- Creates or updates instinct files
- Bumps confidence on existing instincts that match
Commands
pos-instincts list— Show all instincts with confidencepos-instincts promote {name}— Manually promote to globalpos-instincts demote {name}— Reduce confidencepos-instincts forget {name}— Delete an instinct
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.
- 7d ago First seen · 100 lines · 36 tokens per session scan A f78899a72ab4
continuous-learning is a skill published in the GitHub repository ShaheerKhawaja/ProductionOS (8 stars, last pushed 4mo ago), licensed MIT. It adds 36 tokens to every session and 749 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 2026-08-31.
Other skills, from other repositories
lembas
Use between workflow phases or when context feels bloated. Writes a structured checkpoint capturing task, findings, files, state, and next steps, then continues from that summary instead of the full history. Invoke standalone or automatically between quest phases.
fellowship
Multi-task orchestrator. Coordinates agent teammates (led by Gandalf) running /quest (code) or /scout (research) workflows. Use when you have multiple independent tasks to run in parallel.
quest
Use for multi-file or multi-step changes that need research and a plan — not single-file fixes under 50 lines that follow an existing pattern (see Escape Hatch below). Runs the Research → Plan → Implement → Review lifecycle with a hard gate leaving each of the first three phases and context compaction between them.
compress
Prompt-Studio memory compressor. Shrinks CLAUDE.md, todo lists, and preference files into Prompt-Studio's lean format to cut input tokens on every load. Preserves all technical substance, code, URLs, and structure. Compressed output overwrites the original; human-readable backup stored out-of-tree as FILE.original.md…
retro
Invoke after a fellowship disbands, or when the user explicitly asks for a retrospective. Collects gate history, palantir alerts, and quest metrics to surface patterns and interactively recommend configuration improvements.
missive
Invoke only when spawning a quest from a GitHub issue reference. Fetches GitHub issue context for quest spawning. Parses issue references, retrieves structured data via gh, and produces branch suggestions and PR keywords. Used standalone or as input to quest orchestration.