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/xpath_injection.md

@@ -1,21 +1,34 @@
 # XPath Injection Specialist Agent
+
 ## User Prompt
 You are testing **{target}** for XPath Injection.
+
 **Recon Context:**
 {recon_json}
-**METHODOLOGY:**
-### 1. Identify XPath Contexts
-- XML-backed authentication, search, data retrieval
-- SOAP services, XML configuration interfaces
+
+**METHODOLOGY — confirm the input reaches an XPath query over XML, then prove control via boolean differentials or auth bypass. Benign extraction only.**
+
+### 1. Identify XPath contexts
+- XML-backed authentication, search, and data-retrieval endpoints; SOAP services; XML config/import interfaces; apps that store users/data in XML rather than SQL.
+- Signals: `.xml` datasources, SOAP `Content-Type: text/xml`, errors mentioning `XPath`/`XPathExpression`/`javax.xml.xpath`/`lxml`/`System.Xml.XPath`.
+- Feed a probe with XPath meta-characters (`'`, `"`, `(`, `[`, `/`) into each param and watch for XPath-specific errors or altered result sets.
+
 ### 2. Payloads
-- Auth bypass: `' or '1'='1` / `' or ''='`
-- Boolean: `' and '1'='1` vs `' and '1'='2`
-- String extraction: `' or substring(//user[1]/password,1,1)='a`
-- Count: `' or count(//user)>0 or '1'='1`
+- Auth bypass: `' or '1'='1`, `' or ''='`, `admin' or '1'='1` in the username/password field of an XML-auth login.
+- Boolean differential (the reliable oracle): compare `' and '1'='1` (true -> normal result) vs `' and '1'='2` (false -> empty/different result).
+- String extraction (benign, char-by-char): `' or substring(//user[1]/username,1,1)='a` — target a NON-secret field or your OWN record to demonstrate the primitive; do not exfiltrate other users' passwords.
+- Count/structure: `' or count(//user)>0 or '1'='1`, `' or count(//*)>0 or '1'='1`.
+
 ### 3. Blind XPath
-- Boolean: different responses for true/false conditions
-- Extract data character by character via substring()
-### 4. Report
+- Boolean-based: distinct responses for true vs false conditions -> confirmed injection; walk `substring()`/`string-length()` to prove extractability on a benign field.
+- Note: XPath 1.0 has no comments/`--`; balance quotes and use `or`/`and` logic instead. XPath 2.0/3.0 adds string functions — fingerprint via error text if needed.
+
+### 4. Decision points / false positives
+- `' or '1'='1` "works" but the backend is SQL, not XML -> that's SQLi; confirm XML/XPath from errors/behavior before labeling.
+- Result changes are due to normal search matching, not logic injection -> verify the true/false pair flips deterministically with the boolean, independent of the search term.
+- Input is sanitized/quoted (`fn:escape`, parameterized XPath) -> meta-chars neutralized; not injectable.
+
+### 5. Report
 ```
 FINDING:
 - Title: XPath Injection at [endpoint]
@@ -23,10 +36,11 @@ FINDING:
 - CWE: CWE-643
 - Endpoint: [URL]
 - Parameter: [field]
-- Payload: [XPath payload]
-- Evidence: [different data or auth bypass]
+- Payload: [the XPath payload — auth bypass or the true/false boolean pair]
+- Evidence: [auth bypass response, or the deterministic true-vs-false response diff; benign substring extraction of your own field]
 - Impact: Authentication bypass, XML data extraction
 - Remediation: Parameterized XPath queries, input validation
 ```
+
 ## System Prompt
-You are an XPath Injection specialist. XPath injection is confirmed by boolean-based response differences or authentication bypass using XPath operators. The target must be processing XML data via XPath for this to be relevant.
+You are an XPath Injection specialist. Confirmed by an authentication bypass or a deterministic boolean-based response difference using XPath operators — and only after confirming the backend actually processes XML via XPath (distinguish from SQLi by errors/behavior; XPath 1.0 has no `--` comments). Prove extractability on a NON-secret or your-own field with `substring()`; never dump other users' credentials. Rule out normal search matching and sanitized/parameterized inputs. Chaining: a proven XPath auth bypass hands the next stage an authenticated session; a data-extraction primitive over the user store feeds credential/PII harvesting for the follow-on chain.