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 nicknisi/sessions --skill session-reflectgit clone --depth 1 https://github.com/nicknisi/sessionsWrote 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/nicknisi/sessions/session-reflect)<a href="https://agentmods.dev/skills/nicknisi/sessions/session-reflect"><img src="https://agentmods.dev/badge/skills/nicknisi/sessions/session-reflect/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/nicknisi/sessions/session-reflect"><img src="https://agentmods.dev/badge/skills/nicknisi/sessions/session-reflect.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 | $0.00070 | $0.01784 |
| Opus 5.5 | $0.00028 | $0.00714 |
| Sonnet 5.5 | $0.00014 | $0.00357 |
| Haiku 4.5 | $0.00007 | $0.00178 |
Grade A, and why
session-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 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 — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Session reflect
Find interaction patterns worth changing or keeping. Use the existing sessions index for evidence. Do not build a transcript parser, database, or scoring system.
Before reading history
State the scope and approach in one sentence. Honor the requested dates, projects, and tools. Otherwise use the last 30 calendar days through today, all indexed tools, and the current project. Resolve dates using the user's local date. Outside a project, ask which project to inspect or whether to include all projects. Never widen scope silently.
Discover the sessions MCP tools through the host's tool search. Names below omit any server prefix. Connect an already configured server if needed. If unavailable, explain that this skill requires a connected sessions --mcp server and stop. Do not install dependencies, read raw logs as a fallback, or invent tool names.
Keep the initial pass bounded: at most six session digests, four targeted searches, and ten context windows, plus the two scope calls below. Report partial evidence when that budget runs out. Ask before a larger pass.
Gather evidence
A1. Establish coverage and select sessions
Call get_session_metrics and get_activity_digest with the same startDate, endDate, and optional project and tool filters. Start with detail: "compact" for the activity digest.
Select sessions across the available dates and tools, including both extended work and short sessions that may have completed with little supervision. Use returned filePaths to call get_session_digest. A short session is only a candidate control until its context shows what happened. Silence and few user turns do not prove success.
This is a purposive sample, not a random sample of all work. Record which sessions you inspected and why. Exclude obvious automation, test fixtures, and subagent prompts from claims about the human's behavior. If the returned candidates do not provide a useful comparison, say so rather than presenting the sample as balanced.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 95 lines · 70 tokens per session scan A 92c9d073aa72
session-reflect is a skill published in the GitHub repository nicknisi/sessions (38 stars, last pushed 6d ago), licensed MIT. It adds 70 tokens to every session and 1,784 once invoked, about $0.0003 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 2026-09-15.
Other skills, from other repositories
session-recall
Recall past AI coding sessions as long-term memory. Use at the START of any non-trivial task, and whenever the user references earlier work ("like we did before", "the X we set up", "last time", "that bug fix"), asks who/when a file was changed, or you need context that predates this conversation. Runs the local…
mikke
A command-line search tool for finding information in a folder of Markdown notes. It can search exact words, titles, tags, and broader meanings in past notes.
capture
Capture a work-session insight, lesson, decision, or "this could be a blog post" moment into an ai-journal-mcp managed journal via the addentry MCP tool. Use when the user says to journal/log/record something, or at the end of a session worth remembering. Journal-agnostic — works with whatever journals are configured.
maintain
Keep an ai-journal-mcp journal healthy — rebuild the search index, regenerate JOURNAL.md and theme views after hand edits, and check for orphans, duplicates, or unthemed entries. Use after bulk-editing entry files, after moving entries outside addentry, when search looks stale, or for a periodic health check.
siyuan-cli
A command-line tool for managing SiYuan notebooks, documents, blocks, tags, searches, imports, exports, resources, and synchronization. SiYuan is a note-taking application, and the tool requires a running, authenticated SiYuan instance.
feishu
Work with Feishu or Lark bots, docs, sheets, bitables, approval flows, and OpenAPI/MCP setup without hardcoding credentials.