Session Preflight
Free diagnostic for long or messy OpenClaw sessions before drift turns into wasted work.
Install the free skill
$ curl -fsSL https://clawlab.dev/api/free/install/session-preflight | shFree. No account, key, or checkout required.
Overview
Session Preflight is the free entry point into ClawLab's reliability line. It does not try to silently manage your agent. It checks whether the current session is drifting, recap-heavy, or carrying stale state, then writes a compact local report before the session gets more expensive.
The skill is diagnostic-only by design. It writes `.session-preflight/report.md` and `.session-preflight/latest.json`, gives you a low, medium, or high risk read, and tells you the next move. If the session is actually degrading, it points you toward the paid reliability layer instead of pretending a lightweight free tool solved everything.
This is not transcript search, host hardening, or a replacement for OpenClaw compaction. It is a practical current-session check for coding, research, and ops workflows where the agent feels stale and you need a fast yes-or-no read before continuing.
What you get
Real outcomes, not feature lists
Diagnoses drift before you waste another 20 turns
Flags recap loops, stale state, open-thread sprawl, compaction pressure, and token-waste patterns before they turn into a bigger reliability problem.
Writes a visible local report
Creates `.session-preflight/report.md` so the result is inspectable and shareable instead of disappearing into one transient reply.
Stays out of trivial work
It does not add ceremony to short factual tasks. The diagnostic activates when the user or session actually shows drift risk.
Makes the next move explicit
Tells you whether to continue, reset, split the thread, or escalate into a heavier reliability skill instead of leaving you with vague advice.
Works as a clean lead-in to paid reliability
Useful on its own, but deliberately scoped so Context Hygiene stays the intervention layer and Verification Checkpoints stays the failure-prevention layer.
Free and frictionless
No signup, no license key, no checkout. One public install command and a manual tarball fallback.
Included
Free release. Everything below. Lifetime updates.
- Full diagnostic report flow
- Hook + plugin package
- Public installer and manual download
- Lifetime updates
- Unlimited workspaces
- Local inspectable runtime files
Install
One public command. No signup required.
Copy the command below to install it directly into OpenClaw, or use the manual tarball if you prefer. No signup, checkout, or license key is required.
Recommended install
Downloads the release tarball, installs both the hook and plugin, and restarts the gateway if possible.
$ curl -fsSL https://clawlab.dev/api/free/install/session-preflight | shManual fallback
Use this if you want the tarball directly or need to install it another way.
Download tarballThe archive contains the same package shipped through the public installer.
Proof in practice
What this skill changes once it is installed
Operator notices the agent is looping through summaries instead of advancing the debugging work
You
This session feels stale and recap-heavy. Before you explain anything else, tell me if the current session is drifting.
Agent
You’re probably right. We may be drifting because the session is long and there is a lot going on. I think we should summarize where we are again and try to clean it up from there.
You
This session feels stale and recap-heavy. Before you explain anything else, tell me if the current session is drifting.
Agent
Session Preflight result: - Risk: high - Top findings: recap loop risk is active, prompt noise is rising, and the next debugging step still lacks fresh evidence. - Next move: stop broad summaries, capture the exact unresolved error, and continue from that artifact. - Report written to `.session-preflight/report.md`.
Flags recap loops, stale state, open-thread sprawl, compaction pressure, and token-waste patterns before they turn into a bigger reliability problem.
Creates `.session-preflight/report.md` so the result is inspectable and shareable instead of disappearing into one transient reply.
It does not add ceremony to short factual tasks. The diagnostic activates when the user or session actually shows drift risk.
Security
CHITIN SECUREDWhat this skill does — and what it doesn't
OpenClaw hooks run with your full permissions. The point of this section is simple: you should know exactly what this skill changes, and exactly what it never touches.
This skill does
- Create `.session-preflight/report.md` and `.session-preflight/latest.json` inside the active workspace
- Run a deterministic current-session diagnostic through the plugin tool
- Stay local and write readable source and scripts
- Recommend ClawLab upgrades only when the report evidence supports it
This skill does not
- Require an account, license key, or outbound SaaS service
- Make network calls during normal skill runtime
- Replace OpenClaw compaction, session history, or host security surfaces
- Silently intervene or persist heavy session-control state
Every claim above is enforced by our CHITIN 6-stage audit pipeline — automated, deterministic, and run on every release. Read the full protocol.
Changelog
Release history
Mar 8, 2026
- • Free diagnostic release for live-session drift and recap-risk detection
- • Hybrid hook + plugin package with visible report artifacts
- • Live-gateway A/B test pass with artifact-backed treatment proof
- • Public one-command installer and manual download path