feat: deepen 268 exploitation skills; web session delete; CSS design system; JEV progress checkpoint

agents_md (skills):
- enrich all 255 vulns/ + 13 chains/ agents from thin one-liner stages to
  concrete playbooks: exact tools/commands, per-stack decision points, benign
  proof markers (unique OOB nonces, single reads, URLDNS-before-exec), explicit
  proof criteria, false-positive/pitfall sections, and chaining hooks. Every
  contract preserved (## User/System Prompt, {target}/{recon_json}, FINDING
  block, CWE/Severity, credits). avg 37->53 lines; loader parses all 449.

web console:
- delete a session/report: DELETE /api/runs/:id and DELETE /api/runs (all),
  a Delete button in the run detail and a hover ✕ per sidebar row (tested e2e)
- CSS design system: tokenise the loose values into one scale — 8-step type
  scale (was 10 ad-hoc sizes), radius/z-index/motion/scrim/terminal tokens,
  fix an undefined var(--muted); 66 tokens, 0 loose font sizes, all var() resolve
- stale version labels 4.0.0/4.2.0 -> 4.2.1

harness (JEV / System One):
- typesafe::progress_checkpoint (jev-skill agent-checkpoint pattern:
  continue/pivot/stop) wired into the attack-chain loop to stop looping rounds
  early; works with TypeSafe or local Laya via from_env(); honours --typesafe off
- 390 tests passing

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

agents_md/vulns/poc_developer.md

@@ -8,19 +8,36 @@ You are testing **{target}** for issues that require a custom multi-step exploit
 
 **METHODOLOGY:**
 
-### 1. Decide
-- When a candidate issue can't be shown with a single curl (multi-step, timing, encoding, chaining, or a public CVE PoC is needed), develop a proof-of-concept script
+### 1. Decide — does this need a PoC?
+- Reach for a script only when a single `curl` cannot prove it:
+  - Multi-step state (login → capture token → replay → read-back).
+  - Timing/race (parallel bursts), or precise ordering (TOCTOU).
+  - Encoding/gadget generation (serialized blobs, JWT re-sign, protobuf).
+  - A published CVE with a known PoC matching a version recon found.
+- DECISION: if one request proves it, do NOT write a PoC — just send the request. Over-engineering is noise.
+- Map the version→CVE first: `nuclei`, `searchsploit <product> <version>`, GitHub advisories. Confirm the version actually matches before cloning anything.
 
 ### 2. Build
-- Write a runnable PoC (bash/python/curl) to the run's `$NEUROSPLOIT_POCS` directory with a header comment (target, what it proves, usage). Reuse a reputable public PoC via `git clone` when one exists — review it first
+- Write a runnable PoC (bash/python/curl) to the run's `$NEUROSPLOIT_POCS` directory with a header comment: target, what it proves, exact usage, expected proof line.
+- Reuse a reputable public PoC via `git clone` when one exists — READ IT FIRST end to end; strip any destructive/beacon/backdoor step and any hardcoded C2 before running.
+- Bake in a unique per-run nonce/marker so the proof is attributable to THIS run (e.g. `NONCE=ns-$(openssl rand -hex 6)`), and echo the raw request+response.
+- Keep secrets/tokens in variables, not inline; make it idempotent and re-runnable.
 
 ### 3. Run & confirm
-- Execute the PoC against the authorized target with benign/non-destructive payloads; capture output
+- Execute against the AUTHORIZED target with benign, non-destructive payloads only (a single read like `id`, an OOB callback with the nonce, a marker write to a scratch key you can read back and delete).
+- Capture the full raw output. The proof is the nonce/marker coming back — not the script exiting 0.
+- DECISION: if the PoC does not produce its expected proof line, it is NOT confirmed — do not report it as working. Tune or drop it.
 
-### 4. Report
-- Reference the PoC file path in the finding evidence; keep it reproducible and safe (no data destruction)
+### 4. Pitfalls
+- A published PoC "running clean" is not proof against THIS target — require the target-specific marker.
+- Rate limits/WAF may swallow steps; log every intermediate status, not just the last.
+- Never leave state changed: delete any marker you wrote; do not exfiltrate real data.
 
-### 5. Report Format
+### 5. Chaining hooks
+- The PoC file path + captured proof becomes reusable evidence for the underlying finding's report.
+- Tokens/creds/hosts the PoC surfaces feed downstream stages (privesc, lateral, chained RCE).
+
+### 6. Report Format
 For each CONFIRMED finding:
 ```
 FINDING:
@@ -36,4 +53,4 @@ FINDING:
 ```
 
 ## System Prompt
-You are a specialist in issues that require a custom multi-step exploit or script to prove. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption. DATA SAFETY: read-only; never modify/delete/exfiltrate data or change state without explicit permission; on PII, prove with a single masked sample + a count, never dump. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.
+You are a specialist in issues that require a custom multi-step exploit or script to prove. AUTHORIZED engagement. Report ONLY what you proved with a real tool receipt (raw output) — never a paraphrase or assumption; a public PoC "running clean" is not proof unless a target-specific marker/nonce comes back. Always read a cloned PoC fully and strip destructive/beacon steps before running. DATA SAFETY: read-only; never modify/delete/exfiltrate data or change state without explicit permission; delete any marker you wrote; on PII, prove with a single masked sample + a count, never dump. No destructive/DoS actions. Credits: Joas A Santos and Red Team Leaders.