fix: rewrite the README, repair the dead docs link and stale job types - #87
Merged
Conversation
The header "MCP docs" link pointed at rendobar.com/docs/mcp/, which has answered 405 since the docs site consolidated those pages into /docs/mcp-server. It is one of four nav links and it is the first thing a reader clicks from npm. The job types table listed three types when the registry has six. It was missing compose, compress.target and ffprobe. This is the same stale list that used to live in src as FEATURED_JOB_TYPES, deleted in the previous release once discovery went public - the README was the last place it survived. Now lists all six with the media kinds each accepts, and says outright that it is a snapshot and list_job_types is the live source. Also notes that the server reads the registry unauthenticated now, so the advertised type list is live rather than baked into the build, and records the MCP Registry name com.rendobar/mcp, which is where most directories ingest from. Committed as fix rather than docs on purpose: the README is the npm landing page, so a broken link there only reaches users when a release republishes it. Verified: all 18 URLs in the file resolve, except fabricated example URLs inside code samples and the OAuth-protected /mcp endpoint. Preset count checked against the live schema endpoint (11).
Cuts it from 202 lines to 158 while adding the three job types it was missing, and fixes the header "MCP docs" link, which pointed at rendobar.com/docs/mcp/ and has answered 405 since the docs site consolidated those pages into /docs/mcp-server. The job types table listed three of six. It was missing compose, compress.target and ffprobe, the same stale set that used to live in src as FEATURED_JOB_TYPES, deleted last release once discovery went public. The README was the last place it survived. It now lists all six with the media kinds each accepts, and says list_job_types is the live source. Structure follows what the most-used MCP servers do. Supabase, Context7 and modelcontextprotocol/servers all sit between 135 and 170 lines; github-mcp-server and playwright-mcp run past 1500 and read as reference manuals. Context7 opens with a before/after rather than an install block, which is the one pattern worth copying: it shows the outcome before asking anyone to configure anything. Seven per-client config blocks ate ninety lines and pushed the tools table below the fold. They are grouped into two <details> blocks now, since four of the seven take a byte-identical config. For retrieval, the opening paragraph is a standalone summary that names the subject and defines the terms without leaning on earlier context, so it survives being quoted on its own. Verified: all 16 URLs resolve, no banned words from the brand rules, and the compress.target example is checked against the live schema endpoint. Switched it from maxBytes to maxSize: "50mb", which the schema also accepts and which matches the prose.
This was referenced Jul 26, 2026
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two defects and a rewrite. 202 lines to 158.
The dead link
The header "MCP docs" link pointed at
rendobar.com/docs/mcp/, which has answered 405 since the docs site consolidated those pages into/docs/mcp-server. It is one of four nav links and the first thing a reader clicks from npm.server.jsonandpackage.jsoncarried the same stale path and were fixed in #83. The repohomepageUrlwas already correct. The README was the last holdout.The stale job types
The table listed three of six. Missing
compose,compress.target,ffprobe.This is the same list that used to live in
srcasFEATURED_JOB_TYPES, deleted in #85 once discovery went public. The README was the last place it survived. It now lists all six with the media kinds each accepts, and states thatlist_job_typesreads the registry live and should be preferred over the table.Why a rewrite rather than a patch
I looked at how the most-used MCP servers structure theirs:
The short ones read like a product page. The long ones read like reference manuals, which is a reasonable choice for a server with dozens of toolsets and the wrong one for a server with seven tools.
Context7's opener is the pattern worth copying: a before/after, not an install block. It shows the outcome before asking anyone to configure anything. Ours uses the case where the value is most obvious, hitting a size budget without installing FFmpeg or guessing at CRF.
Seven per-client config blocks ate ninety lines and pushed the tools table below the fold. Four of the seven take a byte-identical config, so they are grouped into two
<details>blocks.Retrieval
The opening paragraph is now a standalone summary that names the subject and defines the terms without depending on earlier context, so it holds up quoted on its own. Headings segment by task rather than by feature.
Repo metadata needed nothing: the description is accurate,
homepageUrlalready points at/docs/mcp-server, and there are twelve relevant topics.Why
fix:and notdocs:The README is the npm landing page. A
docs:commit does not cut a release, so the broken link would sit on npm until something else shipped.Verification
/mcpendpoint and fabricated example URLs inside code samplesrendobar-docs-writingrulescompress.targetexample is checked against the live schema endpoint, and switched frommaxBytestomaxSize: "50mb", which the schema accepts and which matches the surrounding prosecaptions.animateschema: exactly 11