Platform-by-platform profile and post audits

Reviewed: 2026-10-11 · Scope: Issue #3 and developer growth. These are offline, user-supplied audit workflows. They do not log in, inspect private accounts, scrape social networks, change profiles, generate fake engagement or publish anything.

One workflow per network, two independent scorecards

Profile audit — completeness/credibility/identity/discovery of a particular account. It uses the specific profile fields documented or provisionally hypothesized for that network. The output includes observations, unknowns, a prioritized fix list, topical advice and the evidence source. A score is reported only when at least 60% of the applicable fields were actually assessed. It is a heuristic percentage of observed present fields, not a platform algorithm, quality grade or predicted reach.
Post audit — evaluate the manually reviewed hook, proof, originality, CTA, accessibility and community requirements of each post, plus platform-native performance numbers if the operator supplied snapshots. Rates are compared only for the same account/platform/format/source and compatible age after publication. Fewer than three comparable posts per format means the median is unknown.
Plan human changes, never automatic optimization of accounts. Write a relevant bio/headline suggestion, draft a better first-frame/hook, schedule a manual experiment, or visit native analytics. Different platforms may require different account roles, eligibility and moderation policies.
Measure, with observations at +24 hours, +7 days and optionally +28 days using each platform’s definitions, and track meaningful external outcomes when available. Views, impressions, reach, Shorts engaged views, interactions and GitHub traffic are different metrics, not additive or interchangeable.
Re-audit, preserve before/after evidence, and report confidence, not a viral-growth guarantee. A post’s popularity is not proof that one copywriting rule caused it.

Platform-specific profile and content strategies

The machine-readable source of truth is reachweaver/profile_audit_registry.json. Profile fields below are recommendation checks, not a claim that every account tier has every feature. Where an official link is unavailable, the whole row is an editorial hypothesis needing live provider verification.

Platform	Profile details to audit	Initial content experiment	Native measurement and caution
LinkedIn	Authentic headshot, clear role/keywords in headline, credible About, work sample, website, relevant banner	Original design decision/diagram with real trade-offs; short technical video or evidence-backed document	Reach/impressions, conversation quality, link clicks; distinguish personal from Page account and eligible Featured features
DEV Community (DEV.to)	About, avatar, relevant tags, technical articles, portfolio link	Detailed original reproducible walkthrough and helpful code snippets	Article views, quality comments and downstream repo visits; editorial standards and transparent AI assistance
YouTube	Banner, channel About/handle, trailer, sections and useful playlists	Real software demo; searchable accurate title/thumbnail and compelling first 30 seconds; separately trial Shorts	Traffic sources, CTR and retention for eligible videos; publish time isn’t known to affect long-term video performance
X	Concise bio (length limit not curl-verifiable on 2026-10-15; see below), header, one destination link, pinned proof	Standalone engineering lesson, screenshot, useful follow-up thread	Impressions/reposts/clicks; no magical hashtag/engagement weights; require original discussion
Bluesky	Honest bio and avatar, website/domain handle when authorized, context-rich pinned post, relevant starter packs	Reproducible observation that belongs in a specific developer custom feed	Reposts and genuine discussion where observable; discovery depends on multiple user-selected feeds
Mastodon	Instance-aware bio (default 500 characters; servers can change it), link verification/profile metadata, accessible header/avatar	Useful technical explanation, relevant hashtags and alt text; ask knowledgeable community members	Boosts, favorites and replies; instance metrics differ; do not invent global impressions
Threads	Clear identity/bio, relevant topic signals, useful pinned context where eligible	Short interesting idea and authentic question/reply	Views/replies measured by native insights when available; avoid engagement bait
Reddit	Community-specific history and rules, authentic profile, disclosure when referencing owned projects	Answer a real subreddit problem thoroughly; owner writes/submits manually	Meaningful comments/upvotes and approved referrals; moderation/rules outrank reach advice, no bulk publishing
Hacker News	Profile About/link and genuine history; know submission guidelines	Human-written, human-posted only original technical story, when appropriate	Constructive discussion only; no AI-generated or AI-edited HN submission text
TikTok	Clear creator identity, original samples, profile CTA as account eligibility permits	Visual before/after demo, clear captions and retention-first structure	Watch behavior, shares, qualified profile actions; no single magic posting hour
Instagram	Bio, avatar, relevant links, Highlights, original pinned examples and category if eligible	Original Reels and multi-step teaching carousels with readable overlays	Native reach, shares, saves, retention; check Professional Dashboard’s updated guidance
Facebook Pages	Business contact/website, accurate description, CTA, cover and avatar	Original demonstration or case-study Reel; real community discussion	Native reach, shares, qualified actions; Meta explicitly deprioritizes trivial duplicate uploads
Telegram	Accurate channel title/photo/description, consented invite, pinned guide	Concise opt-in updates that members genuinely forward	Views and forwarding; channel reach is subscriber-led, NOT an Instagram-style For You algorithm
Discord	Clearly moderated welcome/resources and member permissions	Human community help and release walkthrough	Genuine questions and member retention; do not auto-send promotional messages
Product Hunt	Real maker identity, demo, truthful description, relevant product assets	Substantial product launch, accessible video, meaningful maker comment	Genuine feedback/conversion; do not solicit votes
Pinterest	Searchable boards and profile, connected owned site, original visual identity	Evergreen diagrams, how-to graphics and properly explained destination pages	Impressions, saves, outbound actions; claim owned website when eligible
Hashnode	Complete author page, useful portfolio, readable articles, canonical links	Original technical tutorial; compare search discoverability to DEV	Reader feedback and qualified site visits; scheduling/account availability varies by plan
Indie Hackers	Genuine founder/project details and relevant history	Specific build decisions/lessons from real experiments	Relevant feedback and collaboration inquiries; human-only participation
Peerlist	Project portfolio, clear expertise and links	Original verified project-update case study	Genuine builder connections and referrals; scheduling not verified
Bilibili	Chinese-language channel positioning and original video samples	Localized technical demo with subtitles and provenance	View/retention metrics only as owner export; account eligibility and official ranking details not verified
Zhihu	Chinese bio, credible professional identity and answer portfolio	Evidence-rich technical answer to a real question	Meaningful comments and follow-up; platform ranking specifics not verified
OSCHINA	Engineer profile, article portfolio and source-code links	Practical Chinese-language implementation walkthrough	Developer feedback/conversions; official ranking weights not verified
GitHub	Pinned repos, concise README, correct About/topics, quickstart, contributor guide	Useful actual release, architecture guide, reproducible example	Qualified traffic and accepted contributions; GitHub is not a generic social impression feed
Snapchat	Clear creator identity and proven mobile-native content	Original short vertical demo, tested with a defined audience	Native views/shares when exported; suitability for technical B2B is hypothesis
WhatsApp Channels	Owner-controlled opt-in identity and accurate channel description	Permissioned high-value subscriber announcements	Forwarding/retention only where available; no unsolicited DMs or mass invitations
Google Search (Search profile and Search Console)	Claim your Google Search profile where eligible (`search_profile_claimed`); check Search Console platform properties for your TikTok, Instagram, X and YouTube accounts (`search_console_platform_properties`); website linked from the profile. Each check carries official sources (`check_sources`, checked 2026-10-21)	Not a posting channel: look at how your existing posts and pages are found on Google	Search Console clicks and impressions per platform property; Search profiles are limited to some countries and follower counts (see AI answer engines and search visibility), and not qualifying is not a fault

Primary documentation and evidence dates are saved per platform in the registry. reachweaver registry-age --registry profile-audit reports their age (fresh, aging after 60 days, stale after 120 days) and lists the eight entries without an official source or date as undated (policy). High-confidence capability facts, not undocumented secret weights, include:

LinkedIn profile intro and Page best practices.
YouTube channel customization, retention analysis and publish-time FAQ.
TikTok content recommendation; Instagram Professional Dashboard; Facebook original content.
X profile fields (behind a bot challenge for curl; its 160-character bio limit was read in a real browser on 2026-10-16); Bluesky custom feeds; Mastodon default bio/profile metadata.
DEV editor, analytics and article scheduling; Pinterest claimed website; Product Hunt launch guide, including its prohibition on soliciting upvotes.
Google Search profiles, creating a Search profile (eligibility) and Search Console platform properties, all checked 2026-10-21.
Reddit policies and Hacker News submission guidelines. Neither permits the generic bulk automation playbook.

Sourced profile field limits

profile-audit checks a text field’s length only when profile_field_limits in the registry holds a sourced limit: the official URL, the checked date and a short verbatim quote that was confirmed with curl -sL URL | grep -o '<quote>' on 2026-10-15. The loader refuses an entry without a https:// URL, an ISO date, or a quote that contains the number. A field over its limit raises an advisory length_review alert (with the field, measured length, limit, unit, scope and source). Each report also carries field_limits, which shows every audited text field and the handle as either its sourced limit or "unknown". An unknown limit is never guessed and never checked. API and protocol-schema limits are labeled as such: they describe what the API or record accepts, which can differ from the app editor.

16 sourced limits on 9 of 26 networks (all checked 2026-10-15):

Network	Field	Max	Scope	Source
Instagram	bio	150 characters	official help	Add a bio
Facebook Page	bio (`about`)	100 characters	API reference	Graph API Page
X	display name	50 characters (“typically”)	API reference	X API data dictionary
X	handle	15 characters (“typically”)	API reference	X API data dictionary
X	bio	160 characters	official help (read in a browser 2026-10-16; `curl` gets a bot challenge)	How to customize your profile
YouTube	about (channel description)	1000 characters	API reference	Data API channels
YouTube	handle	30 characters (Latin script)	official help	YouTube handles
Bluesky	display name	64 graphemes	protocol schema	app.bsky.actor.profile
Bluesky	bio (description)	256 graphemes	protocol schema	app.bsky.actor.profile
Mastodon	display name	40 characters (server default)	release source v4.7.3	account.rb; the user guide still says 30
Mastodon	bio	500 characters (server default)	official docs	Setting up your profile
Mastodon	handle (local part)	30 characters (server default)	release source v4.7.3	account.rb
DEV	display name	100 characters	open-source code (Forem)	user.rb
DEV	bio (summary)	200 characters	open-source code (Forem)	profile_validator.rb
DEV	handle	30 characters	open-source code (Forem)	user.rb
Pinterest	bio (About)	500 characters	official help	Edit your profile
GitHub	bio (personal account)	160 characters	official docs	Personalize your profile

Tried on 2026-10-15 and left unknown: LinkedIn headline/About (the server-rendered LinkedIn Help pages found state no number); Threads bio (the help page states no number); TikTok (support.tiktok.com redirects to a JavaScript-only app); Reddit and Discord (Help Centers return a Cloudflare challenge); Snapchat (business help is JavaScript-rendered); Telegram (the API config documents a 70-byte user bio, but the registry audits a channel, so it is not applied); Hashnode, Product Hunt, WhatsApp Channels, Peerlist, Indie Hackers, Bilibili, Zhihu, OSCHINA and Hacker News (no official page with a number was found). Third-party blogs are never used.

Optional owner-authorized native CSV observations

To avoid hand-editing large JSON post lists, supply a normalized, local, owner-authorized post metrics CSV. Both post-audit and suggest can merge it in memory (never a network call) when passed --metrics-csv PATH --owner-authorized. New rows must match the platform/account already present in the profile JSON and may not duplicate an existing post. A blank numeric cell means unknown. More details, example and the exact header contract: PROFILE-METRICS-CSV.md. This is not a verified private API connection or a social algorithm model.

Input file and safe commands

Start with the credential-free illustrative snapshot: examples/profile-post-audit.example.json. Replace sample IDs and fields with information from your own, permitted account, without credentials, emails, cookies or private audience data. Keep real exports in state/ (gitignored), NOT in GitHub.

# These only read an existing local JSON file. No state DB or network call.
reachweaver profile-audit --file examples/profile-post-audit.example.json
reachweaver profile-audit --file state/my-social-snapshot.json --platform linkedin --json
reachweaver post-audit --file state/my-social-snapshot.json --age-hours 168 --tolerance-hours 24 --min-posts 3 --json

The snapshot uses schema_version: "reachweaver.profile-post-audit.v1", a list of profiles, optional brand_name/preferred_handle, and optional posts. Each profile has platform, account_id, handle, optional profile_kind, fields (platform rules), and optional keywords. All field values may be null, meaning not observed; explicit false means the owner actually checked and found the feature missing. Profile completion is not calculated from missing fields. For organizations, handle/name deviations from the declared brand are advisory, not score penalties.

Profile snapshot as CSV

profile-audit --file also reads a profile CSV (a .csv file name), for owners who keep their profile facts in a spreadsheet: examples/profile-snapshot.example.csv. It is long format, one row per field: columns platform,account_id,handle,profile_kind,keywords,field,value, plus optional brand_name and preferred_handle (the same on every row where filled in).

Rows with the same platform and account_id build one profile. handle, profile_kind and keywords (separated by ;) may be repeated or left blank on later rows, but a later row can never contradict an earlier one.
field must be one of the profile fields in the registry. A blank value is unknown, never false and never an empty string; boolean fields accept only true, false or blank. A blank field declares a profile with no fields yet.
Posts are not read from CSV: keep them in the JSON snapshot, and use --metrics-csv (PROFILE-METRICS-CSV.md) for native metrics.
The file goes through the same validation as the JSON snapshot (size limit, credential-like names and values refused, unknown fields refused). Errors name a row number and never print the row. No network is used.

Cross-platform consistency

profile-audit also compares the profiles of one profile_kind across platforms and reports an advisory cross_platform_consistency block (and a short summary in the text output): whether display names, handle stems and website hosts agree, and whether any two bios, headlines or about texts share at least one significant word. Handles and bios legitimately differ by platform, so every finding is a prompt to confirm intent. It never changes a score. Fields left unknown are skipped and listed under not_compared; nothing is guessed. The comparison uses every profile in the snapshot even when --platform limits the per-platform reports.

An optional post observation has platform, account_id, post_id, format, source (the exact owner’s native export label), published_at, observed_at (offset-aware ISO 8601), metrics and review_flags (explicit booleans for hook, proof, originality, CTA, captions, alt text and policy review). The CLI never guesses these qualitative flags from a model. It never contacts an account to verify the fields.

The post report returns a per-post fix list and a descriptive action/exposure rate, when that particular platform/source exports both comparable native metrics. Rates are grouped only for the same platform + account + format + source, at matching ages. The command does not predict expected views and does not rank accounts using synthetic weights. Some platforms do not export both metrics, resulting in insufficient_data rather than a fake zero.

Audit-based improvement procedure

For each platform, produce one human-reviewed action sheet:

Field	Example
Evidence	Owner-provided bio and native impressions; official URL supporting actual profile fields
Profile weaknesses	Missing featured technical proof; no clear action for visitors; unverified account eligibility
One first improvement	Replace vague professional headline with honest expertise + concrete project outcome
Content hypothesis	One original screencast vs one structured technical document explaining the same verified feature
Exposure metric	Native platform impressions or views — use the same exact definition within an experiment
Action metric	Qualified clicks or shares, depending on documented availability
Publishing	Native manual handoff unless a provider adapter, credentials, scopes and exact approval have been verified
Retest	+24 h / +7 d / +28 d; matched topic/format/age; no success claim without source

For actual post text, use references/creative-playbook.md and references/platform-playbooks.md with fresh account context. Do not manufacture testimonials, inflate follower counts, generate fake results, automate engagement or evade moderation. An audit should prioritize what to test next, not promise a particular number of likes.

Known gaps

No automatic profile fetch or account login. Public URLs/screenshots may help a human audit manually, but this command never reads a live profile or configures accounts.
No long-term causal attribution. Observational creative flags and median rates cannot prove recommendation ranking or generate statistically sound uplift without experimental controls.
No exact length guarantee. Only the 16 sourced limits on 9 networks listed under Sourced profile field limits are checked; every other field is unknown. Mastodon servers can differ, grapheme counting is approximate, and provider counters vary (for example Mastodon counts each link as 23 characters).
No automated post creation or scheduling. The profile/post audit commands do not call draft, approve, schedule, publish or change any account.
No deep visual or language quality scoring. A person’s judgement is needed for screenshots, creative hooks, tone, community suitability and multilingual nuance.
No universal social exposure denominator. Native impressions, views, reach and follower snapshots are not comparable between platforms or interchangeable.
No credentials or raw private data in output. The loader rejects credential-like fields/values and oversized input; outputs only audit findings and aggregate metrics.

Future slices can add explicit read-only OAuth/native profile connectors with consent, versioned platform-specific media tests, a human-in-the-loop creative review UI, private account native outcome exports and a preregistered controlled test protocol.