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/sairam0424/MindForgeWrote 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/sairam0424/mindforge/context-budget)<a href="https://agentmods.dev/commands/sairam0424/mindforge/context-budget"><img src="https://agentmods.dev/badge/commands/sairam0424/mindforge/context-budget.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 | $0.00032 | $0.00761 |
| Opus 5 | $0.00016 | $0.00380 |
| Sonnet 5 | $0.00006 | $0.00152 |
| Haiku 4.5 | $0.00003 | $0.00076 |
Grade A, and why
context-budget 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 3d 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.
What it actually says
<execution_context> @.mindforge/skills/context-engineering/SKILL.md </execution_context>
-
Measure token costs per item: For each context item:
- Calculate raw token count
- Assess information density (unique facts per 100 tokens)
- Score temporal relevance (how recently referenced)
- Score task relevance (direct bearing on current objective)
- Identify redundancy (same info stated multiple ways)
- Flag stale items (conversation turns > 10 back with no references)
-
Prioritize by relevance: Apply the priority matrix:
- P0 (Critical): Active task instructions, current file contents, error messages
- P1 (High): Architecture context, type definitions, test fixtures
- P2 (Medium): Historical decisions, related file contents, examples
- P3 (Low): General documentation, old conversation turns, verbose outputs
- Score each item 0-10 on task alignment
-
Apply budget allocation (60/20/10/10): Distribute max-tokens:
- 60% — Active working context (current files, task, instructions)
- 20% — Supporting context (types, tests, architecture docs)
- 10% — History and decisions (conversation, rationale)
- 10% — Buffer for tool results and new information
- Adjust ratios based on task type (debugging needs more history)
-
Identify compaction candidates: Find items that can be compressed:
- Long tool outputs that can be summarized (test results → pass/fail counts)
- Verbose file contents where only specific functions matter
- Repeated context that can be deduplicated
- Conversation segments that can be summarized into key decisions
- Large code blocks where only the interface/signature matters
-
Prune low-value items: Remove or compress based on scoring:
- Drop items scoring < 3 on task relevance
- Summarize items scoring 3-5 into single-line references
- Compact items scoring 5-7 to essential excerpts
- Keep items scoring 7+ in full
- Document what was pruned for recovery if needed
-
Verify essentials retained: Post-optimization validation:
- Confirm all P0 items are present in full
- Verify no broken references (file mentioned but content pruned)
- Check that task can still be completed with remaining context
- Ensure error context is preserved for debugging tasks
- Calculate final utilization percentage and headroom
- Output budget report with before/after comparison
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.
- 3d ago First seen · 71 lines · 32 tokens per session scan A 9b6cda47c94e
context-budget is a command published in the GitHub repository sairam0424/MindForge (0 stars, last pushed 3d ago), licensed MIT. It adds 32 tokens to every session and 761 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 commands, from other repositories
autopilot
Autonomous session-orchestration loop with kill-switches (Phase C-1.b — all 10 kill-switches shipped).
setup
Interview me and scaffold my personal chief-of-staff system (context file, memory, task list, daily brief).
brief
Produce my morning brief — what's on today, what's slipping, what to prep, one thing worth my attention.
loops
Weekly review — surface everything in-flight and force a ship / block / kill decision on each.
postmortem
Analyze the last session and extract instinct proposals.
guide
Invokes the project-guide agent as the entry point for the current project: structured status snapshot, prioritised next steps with skill/agent recommendation, and disambiguation for unclear requests.