> ## Documentation Index
> Fetch the complete documentation index at: https://withcoral.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Custom tools: task examples

> See how custom tools change the tokens, cost and time of real agent tasks.

Coral saves the tool-calling code that an agent writes as a custom tool.
The next agent calls that tool and does not write the code again.
This page shows the effect on real tasks.

The agent ran each task twice:

* **Without custom tools.** The agent started cold, with only the provider tools (Slack, Linear, GitHub).
* **With custom tools.** The agent could also call the custom tools that Coral saved from earlier runs.

Across the 34 tasks below, custom tools gave these results:

> [card] −37% tokens
>
> 12.2M → 7.7M

> [card] −31% cost
>
> $9.86 → $6.79

> [card] −29% time
>
> 47 min → 33 min

> [card] 34 / 34 correct
>
> Up from 33 / 34

## Tasks

Each cell shows the result without custom tools → the result with custom tools.
The percentage is the change.
Each task is a summary of the prompt.
Every answer was correct in both runs, except where a task says otherwise.

<table>
  <colgroup>
    <col />

    <col />

    <col />

    <col />
  </colgroup>

  <thead>
    <tr>
      <th>Task</th><th>Tokens</th><th>Cost</th><th>Time</th>
    </tr>
  </thead>

  <tbody>
    <tr><td>In benchmark mode, do running queries see newly planted flag rows, does a replant replace or add rows, and what does a corrupt flag file do?github `github.fileContentsAtRef` `github.pullRequestDetails` `github.searchPullRequests`</td><td>346,520 → 355,907+3%</td><td>\$0.36 → \$0.34−5%</td><td>72 s → 72 s0%</td></tr>
    <tr><td>A reviewer said a new states table gave the wrong rows. Was the table fixed or dropped, in which commit, and who gave the final approval?github `github.pullRequestReviews` `github.searchPullRequests`</td><td>197,929 → 185,666−6%</td><td>\$0.26 → \$0.20−23%</td><td>45 s → 33 s−27%</td></tr>
    <tr><td>A deploy script fails on a second run within one minute. The bug report was closed for another ticket. Which issue should I follow?linear `linear.issueRelations` `linear.searchIssues`</td><td>314,996 → 286,839−9%</td><td>\$0.27 → \$0.25−10%</td><td>44 s → 42 s−3%</td></tr>
    <tr><td>I reported that a deploy script fails when run twice in one minute, and it was closed for another ticket. Which issue should I follow?linear `linear.issueRelations` `linear.searchIssues`</td><td>317,157 → 266,456−16%</td><td>\$0.28 → \$0.22−19%</td><td>42 s → 36 s−15%</td></tr>
    <tr><td>A report that a deploy fails on a rerun within one minute was closed as a duplicate. Which issue is the canonical one?linear `linear.issueRelations` `linear.searchIssues`</td><td>330,393 → 244,283−26%</td><td>\$0.31 → \$0.24−22%</td><td>50 s → 37 s−26%</td></tr>
    <tr><td>A deploy bug (it fails on a rerun within one minute) was filed twice, apart from a similar CI bug. Which issue is the canonical one?linear `linear.issuePlanning` `linear.issueRelations` `linear.searchIssues`</td><td>326,817 → 266,878−18%</td><td>\$0.28 → \$0.29+1%</td><td>69 s → 49 s−29%</td></tr>
    <tr><td>An auth bug let a corrupt token scope give an agent unrestricted access. What was the issue ID, and which pull request fixed it?github+linear `github.searchPullRequests` `linear.issueComments` `linear.issuePlanning` `linear.searchIssues`</td><td>595,074 → 240,775−60%</td><td>\$0.42 → \$0.20−53%</td><td>95 s → 61 s−36%</td></tr>
    <tr><td>Benchmark runs failed when an image tag was re-pushed. Which issue stopped the build trusting mutable tags, and which PR fixed it?github+linear `github.searchPullRequests` `linear.issueComments` `linear.issuePlanning` `linear.searchIssues`</td><td>325,055 → 283,273−13%</td><td>\$0.23 → \$0.26+11%</td><td>63 s → 60 s−5%</td></tr>
    <tr><td>The nightly benchmark digest stopped comparing with only the previous run. What was the issue ID, and which baseline window did it use?github+linear `github.pullRequestReviews` `github.searchPullRequests` `linear.issueComments` `linear.issuePlanning` `linear.searchIssues`</td><td>342,979 → 267,356−22%</td><td>\$0.31 → \$0.25−19%</td><td>50 s → 49 s−2%</td></tr>
    <tr><td>The nightly benchmark digest got its link to the experiment run back. What was the issue ID, and which PR shipped it?github+linear `github.searchPullRequests` `linear.issuePlanning` `linear.searchIssues`</td><td>482,900 → 246,864−49%</td><td>\$0.34 → \$0.21−38%</td><td>61 s → 45 s−26%</td></tr>
    <tr><td>A cleanup ticket made the benchmark load its test cases once. Which issue was at the start of the chain of blockers that held it up?linear `linear.issueRelations` `linear.searchIssues`</td><td>253,122 → 239,451−5%</td><td>\$0.23 → \$0.20−14%</td><td>46 s → 55 s+19%</td></tr>
    <tr><td>In a thread about a benchmark regression, someone mentioned a graph bug. Which ticket tracked the fix, and which PR shipped it?github+linear+slack `github.searchPullRequests` `linear.issuePlanning` `linear.searchIssues` `slack.searchMessages` `slack.threadLinkedWork` `workspace.searchAll`</td><td>602,256 → 351,791−42%</td><td>\$0.46 → \$0.37−19%</td><td>79 s → 78 s0%</td></tr>
    <tr><td>Someone asked in the engineering channel why the cron alerts were noisy and got an answer. Who were the two people?slack `slack.searchMessages` `slack.threadReplies`</td><td>270,707 → 154,637−43%</td><td>\$0.19 → \$0.14−28%</td><td>58 s → 38 s−34%</td></tr>
    <tr><td>Who asked and who answered, in the engineering channel, how the cron alerts were set up and why they were noisy?slack `slack.searchMessages` `slack.threadReplies`</td><td>376,155 → 155,076−59%</td><td>\$0.30 → \$0.14−53%</td><td>53 s → 40 s−25%</td></tr>
    <tr><td>Someone asked if anyone objected to turning off some database metrics to cut monitoring cost. Who were the two people in that exchange?slack `slack.searchMessages` `slack.threadReplies`</td><td>260,764 → 200,611−23%</td><td>\$0.19 → \$0.16−19%</td><td>60 s → 75 s+23%</td></tr>
    <tr><td>Who were the two people who discussed the cloud account for desktop app code signing in the engineering channel?slack `slack.searchMessages` `slack.threadReplies`</td><td>315,449 → 182,634−42%</td><td>\$0.22 → \$0.14−36%</td><td>64 s → 41 s−37%</td></tr>
    <tr><td>One of our team and one customer worked to connect the customer's dashboards for the first time. Who were the two people?slack `slack.searchMessages` `slack.threadReplies`</td><td>310,920 → 189,488−39%</td><td>\$0.28 → \$0.18−37%</td><td>44 s → 75 s+70%</td></tr>
    <tr><td>When we added rating buttons to investigations, what did we decide to rate when an investigation has several follow-up responses?slack `slack.searchMessages` `slack.threadReplies`</td><td>368,112 → 295,645−20%</td><td>\$0.30 → \$0.24−21%</td><td>54 s → 101 s+87%</td></tr>
    <tr><td>A thread proposed to move prompts out of the codebase because deploys were slow. Which option did we choose, and how did it end?github+slack `github.searchPullRequests` `slack.searchMessages` `slack.threadLinkedWork` `workspace.searchAll`</td><td>259,713 → 216,735−17%</td><td>\$0.22 → \$0.19−16%</td><td>50 s → 52 s+2%</td></tr>
    <tr><td>A thread compared two infrastructure-as-code tools before a cloud resource import. What did we decide, and when would we revisit it?slack `slack.searchMessages` `slack.threadReplies` `workspace.searchAll`</td><td>265,311 → 330,849+25%</td><td>\$0.21 → \$0.29+35%</td><td>50 s → 118 s+133%</td></tr>
    <tr><td>A thread complained about the cost of our infrastructure-as-code tool. What did we decide, and which other options came up?slack `slack.searchMessages` `slack.threadReplies` `workspace.searchAll`</td><td>274,821 → 192,666−30%</td><td>\$0.22 → \$0.18−16%</td><td>71 s → 72 s0%</td></tr>
    <tr><td>A thread listed places to keep prompts outside the codebase to avoid slow deploys. Which option was tried, and what was the final call?slack `slack.searchMessages` `slack.threadLinkedWork`</td><td>207,450 → 161,123−22%</td><td>\$0.21 → \$0.15−28%</td><td>41 s → 48 s+17%</td></tr>
    <tr><td>After a complaint about our infrastructure-as-code tool's yearly cost, what did we decide, and which other option did we not choose?linear+slack `linear.searchIssues` `slack.searchMessages` `slack.threadReplies`</td><td>398,130 → 340,896−14%</td><td>\$0.34 → \$0.26−22%</td><td>100 s → 112 s+12%</td></tr>
    <tr><td>Someone asked about self-hosted docs for a customer. Which PR took the work, and did we publish the docs at once or wait?slack `slack.searchMessages` `slack.threadLinkedWork` `slack.threadReplies`</td><td>321,103 → 173,440−46%</td><td>\$0.23 → \$0.18−24%</td><td>56 s → 43 s−23%</td></tr>
    <tr><td>Someone asked for a backend engineer to fix a slow endpoint that blocked mentions. Who took it, and what was their draft PR number?github+slack `github.searchPullRequests` `slack.searchMessages` `slack.threadLinkedWork`</td><td>248,045 → 255,342+3%</td><td>\$0.22 → \$0.19−16%</td><td>37 s → 87 s+133%</td></tr>
    <tr><td>Someone reported a failed release workflow after a change to its push steps. Which pull request fixed it?slack `slack.searchMessages` `slack.threadLinkedWork`</td><td>423,710 → 173,032−59%</td><td>\$0.36 → \$0.13−65%</td><td>78 s → 34 s−56%</td></tr>
    <tr><td>An editor migration needed backend changes to convert saved documents to markdown. Which pull request shipped that fix?github+slack `github.searchPullRequests` `slack.searchMessages` `slack.threadLinkedWork`</td><td>506,996 → 243,002−52%</td><td>\$0.43 → \$0.19−55%</td><td>91 s → 83 s−9%</td></tr>
    <tr><td>The release workflow started to fail during desktop packaging work, and a fix was promised. Which pull request fixed it?github+slack `github.searchPullRequests` `slack.searchMessages` `slack.threadLinkedWork`</td><td>323,958 → 203,697−37%</td><td>\$0.23 → \$0.15−36%</td><td>49 s → 48 s−1%</td></tr>
    <tr><td>In which channel did someone share cheap flights for the team that went to the offsite?slack `slack.searchMessages` `slack.threadReplies`</td><td>268,707 → 188,696−30%</td><td>\$0.26 → \$0.17−36%</td><td>46 s → 89 s+96%</td></tr>
    <tr><td>In which channel did someone complain that a vendor merged its desktop chat app into its coding agent?slack `slack.messagesByUser` `slack.searchMessages`</td><td>309,061 → 170,667−45%</td><td>\$0.26 → \$0.18−29%</td><td>141 s → 61 s−57%</td></tr>
    <tr><td>In which channel did someone offer to clean up suppressed lint warnings?slack `slack.messagesByUser`</td><td>238,641 → 124,227−48%</td><td>\$0.22 → \$0.10−51%</td><td>130 s → 34 s−74%</td></tr>
    <tr><td>In which channel did someone share an AI agent's diagnosis of why our release automation failed?Answer: wrong without custom tools, correct with themslack `slack.messagesByUser`</td><td>1,178,574 → 127,995−89%</td><td>\$0.73 → \$0.11−84%</td><td>516 s → 27 s−95%</td></tr>
    <tr><td>In which channel did someone ask for help with a list of prompts to track if AI assistants recommend us?slack `slack.messagesByUser` `slack.searchMessages` `slack.threadReplies`</td><td>355,140 → 233,702−34%</td><td>\$0.23 → \$0.17−27%</td><td>158 s → 83 s−48%</td></tr>
    <tr><td>In which channel did someone complain that our public example agent repo needed a dependency update?slack `slack.messagesByUser`</td><td>251,346 → 135,489−46%</td><td>\$0.25 → \$0.13−46%</td><td>149 s → 31 s−79%</td></tr>
  </tbody>
</table>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.