@cursor/bdk 0.2.9 → 0.2.10

package/dist/docs/assets/{building-with-agents.md.BPjPFxFg.js → building-with-agents.md.Df87rAe1.js}

@@ -6,4 +6,4 @@ import{_ as a,c as t,o as n,a3 as i}from"./chunks/framework.BNw1pucY.js";const u
 <span class="line"><span>  --message &quot;Is https://github.com/acme/checkout/pull/42 ready to approve?&quot;</span></span></code></pre></div><p><code>run</code> prints the saved trace path. Use that exact path instead of constructing one from the session ID:</p><div class="language-text vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">text</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span>bdk trajectory --events &lt;trace-path-printed-by-run&gt;</span></span>
 <span class="line"><span>bdk eval --dir . --tag smoke</span></span></code></pre></div><p>Finish with the project&#39;s TypeScript check. BDK executes TypeScript through type stripping, so a successful turn does not prove the project type-checks.</p><h2 id="hand-off-credential-dependent-checks" tabindex="-1">Hand off credential-dependent checks <a class="header-anchor" href="#hand-off-credential-dependent-checks" aria-label="Permalink to &quot;Hand off credential-dependent checks&quot;">​</a></h2><p>Project discovery, <code>info</code>, server-tool calls, and serve bring-up do not need a model credential. Model turns and evals resolve <code>CURSOR_API_KEY</code>, <code>CURSOR_API_KEY_FILE</code>, <code>CURSOR_SERVICE_ACCOUNT_KEY</code>, then <code>bdk login</code>.</p><p>When the coding environment has none, it should complete every key-free check and hand you the exact remaining commands:</p><div class="language-text vp-adaptive-theme"><button title="Copy Code" class="copy"></button><span class="lang">text</span><pre class="shiki shiki-themes github-light github-dark vp-code" tabindex="0"><code><span class="line"><span>bdk run --dir . --message &quot;&lt;fixture prompt&gt;&quot;</span></span>
 <span class="line"><span>bdk serve --dir . --dev</span></span>
-<span class="line"><span>bdk eval --dir . --tag smoke</span></span></code></pre></div><p>Missing access is a handoff boundary, not a reason to skip verification or invent a substitute result.</p><h2 id="improve-a-working-agent" tabindex="-1">Improve a working agent <a class="header-anchor" href="#improve-a-working-agent" aria-label="Permalink to &quot;Improve a working agent&quot;">​</a></h2><p>After one smoke turn passes, give the coding agent fixed fixtures, one success criterion, and the behavior that must not change. The hillclimb skill measures the current run, changes one lever, remeasures, and adds an eval for every kept improvement.</p><blockquote><p>Hillclimb the PR triage agent on these two saved PRs. Reduce repeated reads without losing either gold finding. Keep the review JSON shape and all existing evals unchanged.</p></blockquote><p>Ask it to read <a href="./../skills/hillclimb/SKILL.html"><code>skills/hillclimb/SKILL.md</code></a>. Keep creation and improvement separate: a smoke turn proves the surface works; a measured climb proves it works better on fixed inputs.</p><h2 id="related" tabindex="-1">Related <a class="header-anchor" href="#related" aria-label="Permalink to &quot;Related&quot;">​</a></h2><ul><li><a href="./quickstart.html">Quickstart</a>: build the first agent manually</li><li><a href="./evals.html">Evals</a>: protect behavior with regression cases</li><li><a href="./hillclimbing.html">Hillclimbing</a>: run one measured improvement round</li><li><a href="./guides/improve.html">Self-improvement</a>: let a running agent propose a source change through a pull request</li><li><a href="./reference/cli.html">CLI reference</a>: command contracts and flags</li></ul>`,35)])])}const g=a(o,[["render",s]]);export{u as __pageData,g as default};
+<span class="line"><span>bdk eval --dir . --tag smoke</span></span></code></pre></div><p>Missing access is a handoff boundary, not a reason to skip verification or invent a substitute result.</p><h2 id="improve-a-working-agent" tabindex="-1">Improve a working agent <a class="header-anchor" href="#improve-a-working-agent" aria-label="Permalink to &quot;Improve a working agent&quot;">​</a></h2><p>After one smoke turn passes, give the coding agent fixed fixtures, one success criterion, and the behavior that must not change. The hillclimb skill measures the current run, changes one lever, remeasures, and adds an eval for every kept improvement.</p><blockquote><p>Hillclimb the PR triage agent on these two saved PRs. Reduce repeated reads without losing either gold finding. Keep the review JSON shape and all existing evals unchanged.</p></blockquote><p>Ask it to read <a href="./../skills/hillclimb/SKILL.html"><code>skills/hillclimb/SKILL.md</code></a>. Keep creation and improvement separate: a smoke turn proves the surface works; a measured climb proves it works better on fixed inputs.</p><h2 id="related" tabindex="-1">Related <a class="header-anchor" href="#related" aria-label="Permalink to &quot;Related&quot;">​</a></h2><ul><li><a href="./guides/migrate.html">Early-alpha migration</a>: hand an architecture v1 migration to <code>bdk migrate</code> or the bundled coding-agent skill</li><li><a href="./quickstart.html">Quickstart</a>: build the first agent manually</li><li><a href="./evals.html">Evals</a>: protect behavior with regression cases</li><li><a href="./hillclimbing.html">Hillclimbing</a>: run one measured improvement round</li><li><a href="./guides/improve.html">Self-improvement</a>: let a running agent propose a source change through a pull request</li><li><a href="./reference/cli.html">CLI reference</a>: command contracts and flags</li></ul>`,35)])])}const g=a(o,[["render",s]]);export{u as __pageData,g as default};