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 tranhieutt/software_development_department --skill save-stategit clone --depth 1 https://github.com/tranhieutt/software_development_departmentWrote 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/tranhieutt/software_development_department/save-state)<a href="https://agentmods.dev/skills/tranhieutt/software_development_department/save-state"><img src="https://agentmods.dev/badge/skills/tranhieutt/software_development_department/save-state/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tranhieutt/software_development_department/save-state"><img src="https://agentmods.dev/badge/skills/tranhieutt/software_development_department/save-state.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00054 | $0.01361 |
| Opus 5.5 | $0.00022 | $0.00544 |
| Sonnet 5.5 | $0.00011 | $0.00272 |
| Haiku 4.5 | $0.00005 | $0.00136 |
Grade A, and why
save-state 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 1mo 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 — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Dump the current working context into production/session-state/active.md as a session checkpoint, AND write an atomic per-task checkpoint to .tasks/checkpoints/[task_id].md when a task ID is provided.
Session checkpoint is automatically read by session-start.sh at the next session start and surfaced by pre-compact.sh before context compaction.
Per-task checkpoints are used by /resume-from [task_id] to restore granular cognitive state at the exact point of failure.
Steps
1. Parse arguments
Parse $ARGUMENTS to extract:
- task_id (optional): First token matching pattern
^\d{3}or a word without spaces (e.g.042,auth-api). If present, an atomic checkpoint will be written. - note (optional): Remaining text after task_id, used as a note.
2. Gather context
Before writing anything, collect the following from the current conversation and working state:
- Current task: What is the primary task being worked on right now?
- agent_id: Which agent is currently executing (e.g.
backend-developer,qa-engineer)? - Progress: What has been completed in this session? List key milestones.
- Decisions made: What architectural, design, or implementation decisions were made? (These are the most important to preserve.)
- Files modified: Run
git diff --name-only && git diff --staged --name-only && git ls-files --others --exclude-standardto get the current working tree state. - output_snapshot: A concise snapshot of the last significant output — last file written, last test result, last API response shape, or last command output.
- Open questions: What is unresolved or blocked?
- Next step: What is the very next action to take when resuming.
If a note was parsed from $ARGUMENTS, append it to the "Notes" section.
2. Extract Durable Memory (Native System)
Are there any lessons learned, coding patterns established, constraints added, or technical decisions made in this session that apply globally to the project?
If so, before writing active.md, extract them to the .claude/memory/ directory:
- Write or update a specific topic file (e.g.
project_tech_decisions.mdorfeedback_rules.md), ensuring it has the required YAML frontmatter (name,description,type). - Update
.claude/memory/MEMORY.mdwith a pointer to that file if it's not already listed.
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.
- 1mo ago First seen · 159 lines · 54 tokens per session scan A 3cd950620377
save-state is a skill published in the GitHub repository tranhieutt/software_development_department (73 stars, last pushed 5mo ago), licensed MIT. It adds 54 tokens to every session and 1,361 once invoked, about $0.0002 per session on Opus 5.5. 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 skills, from other repositories
context-bankruptcy
Declare bankruptcy on a long-lived AI agent's accumulated memory — audit what it currently believes, separate ground truth from stale and wrong, purge deliberately, restate the truths that survive, and log what was lost. Use when an agent keeps acting on outdated facts, contradicts itself across sessions, 'remembers'…
claude-reflect
Self-learning system that captures corrections during sessions and reminds users to run /reflect to update CLAUDE.md. Use when discussing learnings, corrections, or when the user mentions remembering something for future sessions.
company-brain
Your team's shared, AI-ready knowledge base — people, companies, meetings, SOPs, and decisions structured so an agent can answer on your team's behalf. Team-scope sibling to second-brain. Modes — capture, compile (wiki pages + INDEX.md), query (trust-weighted, saved to outputs/), review (verify / deprecate / supersede…
second-brain
When you want to capture into, compile, query, lint, or connect your personal Second Brain. Wraps the Karpathy LLM Wiki schema (Obsidian or any markdown vault) — raw/ (unprocessed sources), wiki/ (AI-compiled interlinked topic pages), outputs/ (generated artifacts). Tool-agnostic in design but defaults to a vault at…
context-engineering-review
Review what an LLM feature or agent actually puts in its context window — and find what's bloating, missing, or fighting itself. Use when asked to review a system prompt and context assembly, cut token usage without losing quality, debug an agent that ignores instructions, or audit how retrieval results, history, and…
memory-file-maintenance
Keep your AI memory/context file (MEMORY.md, CLAUDE.md, custom instructions) healthy over time — pruning the stale, adding the new, and keeping it sharp so your AI keeps getting you right. Use when asked review my memory file, my AI context is outdated, clean up my CLAUDE.md, or maintain my AI instructions. Produces a…