Extracting the gaussian behavioral timing module from jobs-apply into a standalone npm package and applying it to pirate-ship API calls will reduce provider quarantine events by 50%+ without modifying the core application logic

Hypothesis

Extracting the gaussian behavioral timing module from jobs-apply into a standalone npm package and applying it to pirate-ship API calls will reduce provider quarantine events by 50%+ without modifying the core application logic

Changelog

Date	Summary
2026-04-06	Audited: added Changelog, domain tags, stamped last_audited
2026-04-04	Initial creation

---

Hypothesis

We bet that the behavioral timing module : already proven to prevent LinkedIn account restrictions in the anti-detection suite experiment : is a general-purpose solution applicable to any automation project that makes external API calls. The gaussian behavioral timing research established that human inter-action intervals follow approximately log-normal distributions, not uniform ones. The behavioral detection pitfall showed the consequence: uniform timing signatures are statistically detectable within 20-30 requests. The source idea proposes packaging the Box-Muller sampler as a standalone npm module.

The hypothesis is that extracting this into @jobs-apply/gaussian-timing and applying it to pirate-ship’s provider layer will reduce quarantine events (429 responses, circuit breaker trips) by 50%+ without modifying core application logic, because the timing distribution is the only behavioral signal that needs to change.

Method

1. Extract module: isolate the gaussian timing functions from packages/engine/src/utils/timing.ts into a standalone package. Interface: gaussianDelay(mean, stddev, min, max) returning a Promise that resolves after the sampled delay.
2. Package: publish as @jobs-apply/gaussian-timing with TypeScript types. Export: gaussianDelay, boxMullerSample, GaussianTimingConfig.
3. Pirate-ship integration: replace fixed time.sleep(delay) calls in the provider layer with gaussian equivalents. Target: inter-API-call delays, retry backoffs, and session pacing.
4. Measurement setup: instrument pirate-ship to log provider quarantine events (429 responses, timeout errors) and their timestamps. Establish a 2-week baseline with fixed timing, then a 2-week test period with gaussian timing.
5. Statistical test: compare quarantine event frequency (events per 1000 API calls) between baseline and test periods. Use a two-proportion z-test with alpha=0.05.

Results

Pending. Will measure:

• Provider quarantine events per 1000 API calls (baseline vs gaussian)
• 429 response rate (baseline vs gaussian)
• Session duration variance (should increase with gaussian timing)
• Package size and API surface area

Findings

Pending.

Next Steps

If quarantine reduction is confirmed, apply the same module to redcorsair MCP server (OAuth token refresh timing, tool call pacing). The module could eventually become a vault skill : promoted from experiment to skill when it proves general across 3+ projects (LinkedIn, pirate-ship, redcorsair). The promotion criterion maps directly to the skill sync lifecycle: pattern validated in experiment → promoted to vault skill → auto-synced to mounted SKILL.md.